A Comprehensive Clinical Review of Adult-Type Diffuse Glioma Incorporating the 2021 World Health Organization Classification

The American Society of Neuroradiology (ASNR) is accredited by the Accreditation Council for Continuing Medical Education (ACCME) to provide continuing medical education for physicians. The ASNR designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit^TM. Physicians should claim only the credit commensurate with the extent of their participation in the activity. To obtain Self-Assessment CME (SA-CME) credit for this activity, an online quiz must be successfully completed and submitted. ASNR members may access this quiz at no charge by logging on to eCME at asnr.mycrowdwisdom.com. Nonmembers may pay a small fee to access the quiz and obtain credit via asnr.mycrowdwisdom.com.

Adult-type diffuse gliomas are the most common primary malignant brain tumor, having an incidence of 5–6 per 100, 000 persons annually.1 They represent a heterogeneous group of neoplasms carrying numerous genetic, epigenetic, and transcriptomic aberrations 2 that demonstrate a propensity for proliferation, brain invasion, migration, and, ultimately, resistance to current therapeutic options. Consequently, the diagnosis is one of an incurable disease with overall survival linked to histologic grade and molecular characteristics. Glioblastoma (GBM) is the most malignant diffuse glioma, with patients surviving an average of 12–15 months following diagnosis.3

The management of patients with diffuse glioma requires a multidisciplinary approach incorporating a quorate of oncology, neurosurgery, radiology, histopathology, clinical genetics, and allied health professionals. Each team member needs core knowledge that encompasses an appreciation of normal brain composition and function, fundamental tumoral pathophysiologic mechanisms, histologic and molecular diagnoses, radiologic methods, and treatment strategies. Maintaining contemporary insight into these fields is challenging but is particularly important as research evolves to deepen our understanding of gliomas, evidenced by the recent 2021 changes to the World Health Organization (WHO) Classification of CNS Tumors,4 all of which were driven by a greater understanding of the molecular mechanisms involved in gliomagenesis. Updates in the classification system have resulted in a search for radiologic biomarkers that align with diagnosis and prognosis. Moreover, knowledge of molecular tumor drivers has inspired exploration into new and targeted treatment avenues.

The reader will be taken through a series of relevant subsections beginning with a brief overview of normal brain composition, followed by insights into the cell of origin of diffuse glioma, tumor heterogeneity, the tumor microenvironment, and glioma plasticity. Subsequently, diffuse glioma classification, including the recent WHO Classification of 2021, will be explained with particular reference to key molecular pathways. Reference to prognosis, current treatment regimens, and potential therapeutic targets will be made. These concepts will be underpinned by frequent references to radiology. The intent of this review is not only to provide the clinician with a comprehensive insight into diffuse glioma but also to serve as a clinical connection for nonclinical basic and translational scientific researchers.

Brain tissue comprises 3 broad compartments: the neural cells, the vascular system, and the interstitial system (Fig 1).

Fig 1.Click to view

Fig 1.

Three broad compartments of the brain: the neural cells, the vascular system, and the interstitial system with their respective percentages of total brain volume.150 Neural cells comprise the main functional element of the brain, while the vascular and interstitial systems provide a supportive, living microenvironment. The neurovascular unit includes capillary endothelial cells connected by tight junctions, surrounded by a specialized basement membrane that is shared with pericytes and astrocytic endfeet and sparsely interconnected by neuronal endings and microglia. Together these components dictate the physical properties of the BBB. The ISS, comprising the ISF, is depicted as the white spaces between neural cells and the ECM. The ECM is made up of components created by the cells within it: neurons, astrocytes, oligodendrocytes, and microglia. It has 3 major components: the basement membrane (lying around the blood vessels), the neural interstitial matrix (diffusely distributed), and the perineural nets (a lattice that wraps around neurons creating very close synaptic contacts; it is composed of a hyaluronan backbone, chondroitin sulphate proteoglycans, tenascins, and hyaluronan and proteoglycan link proteins).151 The homeostasis and health of the brain relies on the proper function of the ISS, a function that can be disrupted by the invasion of glioma cells. GPC indicates glial progenitor cells; NG2, oligodendroglial precursor.

 

Neural Cells

The human brain consists of >200 billion neural cells. These cells form complex networks, connected with 15–20 trillion chemical and electrical synapses.5 The 2 main groups of neural cells are neurons and glia.

The neuron, responsible for sending and receiving nerve impulses or signals, consists of a cell body (or soma, housing the nucleus), dendrites (receiving input from other cells), and an axon (the output structure). Most of the neuronal cell bodies of the brain are found in gray matter and have diameters ranging between 5 and 10 µm. Some neurons are confined to the gray matter. Others project axons into the white matter, typically arranging into axonal bundles called neuronal tracts. Historically, these tracts have been ascribed differing nomenclature, for example, commissures or fasciculi, but they have the common function of connecting different regions of the CNS. By way of anatomic arrangement, these tracts inadvertently offer a “highway” for cancer cells to spread.

Glia, also referred to as glial cells and neuroglia, can be divided into 4 major groups: 1) microglia, 2) astrocytes, 3) oligodendrocytes, and 4) their progenitors, neuron-glia antigen 2-expressing glial cells.6 Astrocytes, oligodendrocytes, and NG2-glia are cells of neural origin having originated from the ectoderm. Conversely, microglia originate from yolk sac progenitors and only populate the brain during development.7,8 Glia are far from being the simple neuronal supportive “glue” to which their name refers but instead have a range of functions and form the homeostatic and defensive arm of the nervous system (Fig 2).5

Fig 2.Click to view

Fig 2.

The 4 major groups of glial cells with a synopsis of their main functions and cell markers. GFAP indicates glial fibrillary acidic protein.

 

A long-held hypothesis was that gliomagenesis (diffuse glioma formation) represented neoplastic transformation of fully-differentiated glia.16 Fundamental to this theory were 2 general concepts: first, an appreciation that diffuse gliomas share common features of their normal respective cellular counterparts, including histologic, functional, and molecular characteristics;17 and second, the belief that adult glia were the only dividing cells in the postnatal brain, making them the only brain cells susceptible to transformation.16 The latter assumption was disproven with the discovery of other proliferative populations of cells in the adult human brain. Neural stem cells (NSCs), which are multipotent and self-renewing, have been isolated from various brain locations, including the subventricular zone (lining the lateral ventricles), the dentate gyrus of the hippocampus, and the subcortical white matter. Glial progenitor cells (self-renewing precursors capable of producing astrocytes and oligodendrocytes) have also been found in adult mammalian CNS locations, including the cortex, the corpus callosum, the periventricular white matter, the subventricular zone, and the dentate gyrus of the hippocampus. Thus, neural stem and progenitor cell pools, in addition to differentiated glia, are potential substrates for neoplasia.17 Recently, strong evidence has emerged that that glioblastoma arises from NSCs within the subventricular zone of the brain rather than mature glia.18 At present, however, scientific consensus as to the exact cell of origin (often termed the brain tumor–initiating cell) remains elusive (Fig 3).

Fig 3.Click to view

Fig 3.

Possible tumour initiating cells. Self-renewing, multipotent neural stem cells (NSCs) may differentiate into different fate-restricted progenitors such as neuronal progenitor cells (NPC), astrocyte progenitor cells (APC) and oligodendrocyte progenitor cells (OPC). NSCs, progenitor cells or even differentiated glia could be transformed into diffuse gliomas through genetic and epigenetic changes (shown as red arrows). Figure modified from Lu et al.2

 

Diffuse gliomas can show differences from patient to patient, so-called intertumor heterogeneity. They also demonstrate intratumoral heterogeneity, a concept referring to the unique phenotypic, genetic, and functional differences arising across the landscape of an individual tumor. This heterogeneity enables distinct cancer cell populations to behave and respond in a manner different from environmental factors, including in response to therapeutic strategies. While some cell populations may be successfully eradicated by treatment, others may survive, derive resistance, and cause recurrence.19 Two theoretic models have been postulated to explain tumor development and intratumoural heterogeneity: 1) the clonal evolution model, and 2) the cancer stem cell model.

The clonal evolution theory describes tumorigenesis (or in the context of glioma, gliomagenesis) as a consecutive process of mutational acquisition that promotes clonal expansion and diversification.20 Genetic or epigenetic mutations appear randomly, with any new clones undergoing the pressure of natural selection (such as environmental changes generated through chemo- or radiation therapy). These generate survival of the fittest with such clones expanding and outgrowing others. In this model, heterogeneity exists through the presence of the remaining weaker clones generated during tumor expansion.19

In contrast, the cancer stem cell model describes a situation in which only a subset of cells possess the capacity to sustain tumorigenesis through self-renewal and to generate functional heterogeneity through a hierarchic process of differentiation. Cancer stem cells divide asymmetrically and generate cancer stem cells and more differentiated daughter cells. Heterogeneity is produced through the mixture of a smaller population of cancer stem cells (tumourigenic) and a tumor bulk of noncancer stem cells with various degrees of differentiation (nontumourigenic) and limited proliferation capability.

Cancer stem cells have been discovered in a variety of tumors including gliomas, where they are renamed “glioma stem cells” (GSCs).21 GSCs are a subpopulation of undifferentiated cells that have potent plasticity and can differentiate into astrocytes, oligodendrocytes, and neuronal lineages. These cells also show relative resistance to radiation and chemotherapy and are key enablers of tumor recurrence (Fig 4).17,22

Fig 4.Click to view

Fig 4.

GSCs are tumor-propagating cells that coordinate growth and invasion. Although the exact origin of GSCs is unknown, suggestions have included the following: 1) from transformed NSCs; 2) from dedifferentiation of normal glia; 3) from dedifferentiation of tumor cells; and 4) from oligodendroglial precursors (NG2).17

 

Recently, Lan et al22 studied the clonal evolution of barcoded GBM cells following xenotransplantation to define their individual fate behaviors. They showed that the growth of GBM clones in vivo involves a conserved proliferative hierarchy rooted in GSCs, whereby slow-cycling stem cells give rise to a more rapidly cycling progenitor population with extensive self-maintenance capacity, which, in turn, generates nonproliferative cells.22 The group also identified rare outlier clones that deviated from these tendencies.22 The authors suggested that in common with other cancer models,23,24 heterogeneity in clonal expansion does not derive from genetic diversity but instead emerges as the predictable outcome of chance fate decisions made by GSCs and their progeny.22

Presently, the exact nature of cells that may initiate a glioma and those that maintain and promote growth and invasion are still to be determined. Theories of cell of origin and GSCs are often presented as different entities. The cell of origin concept could be thought of as referring to “tumor-initiating cells,” whereas the glioma stem cell concept may be better considered “tumor-propagating cells.” A single tumor may contain multiple glioma stem cell clones, but they are proposed to share a common ancestor: the cell of origin that acquired the first oncogenic event.

Further complexity of tumor heterogeneity derives from the interaction between the glioma and its microenvironment. Cell plasticity describes the ability of cells to change their phenotypes without genetic mutations in response to environmental cues.25 Different microenvironmental factors (eg, composition of the ECM, differences in oxygen pressure, vessel density, local growth factors) may cause clonal heterogeneity and can allow glioma cells to proliferate, migrate, and invade the brain parenchyma. The next section further examines the tumor microenvironment (TME) and its links to glioma plasticity.

In addition to a heterogeneous population of cancer cells, a glioma is also composed of a variety of resident and infiltrating host cells, secreted factors, and ECM proteins, collectively known as the TME.26 Broadly, the TME has vascular, cellular, metabolic, and biomechanical contributions.13

It is worthwhile to briefly explore the evolution of the classification system to understand the various terminologies used in diffuse glioma nomenclature. This can be appreciated by considering pre-2016, in 2016, and beyond 2016 compared with the 2021 WHO Classification.

Before 2016

For several decades, gliomas were classified according to the system devised by Bailey and Cushing.46 This system was based on histologic phenotypes and their resemblance to glia, with glia largely considered to be the putative cells of origin of the tumors.47 Under a microscope, astrocytomas resemble the morphology of astrocytes, whereas oligodendrogliomas mostly resemble the morphology of oligodendrocytes and mixed oligoastrocytoma resemble a cellular combination. As tumors become more aggressive, the cell morphology changes and appears less differentiated. The presence of histopathologic features like atypia, anaplasia, mitotic activity, microvascular proliferation, and necrosis defines the grade of the diffuse glioma, ranging from II to IV, with higher grades corresponding to poorer prognosis. Certain radiologic markers permitted an estimate of the underlying histology and grade (Fig 5).

Fig 5.Click to view

Fig 5.

Historic WHO Classification of diffuse glioma based on histologic resemblance. Diffuse gliomas are graded according to morphologic changes, grades 2, 3, and 4. Aggression of tumor and prognosis worsens with increasing grade. Grade 2 tumors are also designated LGG, and grade 3 and 4 tumors are grouped as HGG. In the clinical setting, diffuse gliomas are preferentially assessed with MRI owing to their intrinsically high-soft-tissue contrast definition. Occasionally, CT is used when MRI is contraindicated (eg, in patients with MRI-incompatible cardiac or neurologic devises, severe claustrophobia). The international Standardized Brain Tumor Imaging Protocol has established the minimum image acquisition requirements for MRI.152A–E, Grade 2 (LGG) oligodendroglioma. Classic radiologic features of an oligodendroglioma show a cortical-/subcortical-based calcific mass (A, Foci of high density on axial CT), with ill-defined borders (B, Axial T2-weighted MRI), heterogeneous intrinsic T1 signal (C, Axial T1-weighted MRI), and patchy foci of enhancement (D, Axial T1-weighted MRI post-T1Gd). Histology reveals typical round cells with blue nuclei (E). F–J, Grade 2 (LGG) astrocytoma. A well-demarcated tumor devoid of calcium (F, Axial gradient-echo MRI) with intrinsic high T2 signal (G) and showing more homogeneous signal on T1 (H) and no abnormal contrast enhancement (I). Histology reveals increased cellularity and pleomorphism (J). K–N, Gliomatosis cerebri. A diffusely infiltrative WHO grade 3 glioma shown as high signal on FLAIR MRI sequences (K–N). The tumor has infiltrated into the gray and white matter of all lobes in both cerebral hemispheres, especially noticeable in the splenium of the corpus callosum (M) and medial aspect of the right temporal lobe (N). These tumors did not show enhancement. The presence of enhancement would suggest multifocal glioblastoma. The histologic diagnosis of gliomatosis cerebri was dropped in the 2016 WHO classification. The term can still be used as a radiologic description of tumor. O, Histology of a diffuse glioma (right side of the slide) shows greater cellularity and pleomorphism than the adjacent brain (left side of the slide), but the margin is indistinct, illustrating the infiltrative nature of the tumor (O). P–T, Grade 4 GBM, the most malignant diffuse glioma. Typical radiologic appearances of a centrally necrotic (P, DWI), peripherally enhancing mass (Q and R), with surrounding and infiltrative high T2 signal (S). Histology (T) shows nuclear atypia, increased proliferation, microvascular hyperplasia (arrow), and necrotic foci (asterisk), typically surrounded by “pseudopalisading” cells (arrowheads). The historic term “glioblastoma multiforme” was relinquished in the 2016 WHO Classification, preferring the simpler glioblastoma. The terms low-grade (WHO 2) and high-grade (WHO 3 and 4) glioma are presently used in the clinical setting. The current best radiologic marker to differentiate LGG from HGG is the presence of tumoral enhancement on T1Gd imaging. This is not always reliable for oligodendrogliomas because intrinsic tumor enhancement resulting from their inherent hypervascularly can mimic a higher-grade tumor on imaging grounds. Additionally, while non-enhancement generally suggests LGG, approximately 30%–40% of nonenhancing gliomas are, in fact, malignant.153 Due to the intratumoural heterogeneity, these high-grade components can actually coexist with low-grade tumor within the same nonenhancing lesion. Lack of a T1Gd-enhancing target makes it difficult to reliably localize those high-grade components using conventional MRI alone.154 These factors can contribute to a reported 30% incidence of sampling error and misdiagnosis (ie, undergrading) of nonenhancing high-grade gliomas.155

 

Historically grades were numbered using Roman numerals, but this has recently changed to Arabic nomenclature to prevent interpretation error (ie, 2 versus 3 is less susceptible to error in a report than II versus III). The Arabic terms will be used herein.48

This classification system, used until 2016, had high interobserver variability among pathologists, and survival varied substantially within grades.3,47,49 This issue led researchers to explore markers that would improve the characterization of clinically relevant subgroups.

2016

The discovery of key genetic mutations in diffuse gliomas that conferred varying prognostic profiles triggered the production of the 2016 World Health Organization Classification of Central Nervous System Tumors, which integrated phenotypic and genotypic parameters.47 This classification focuses on the presence or absence of 2 principal tumor markers: 1) mutations in IDH1 or IDH2 (encoding isocitrate dehydrogenase cytoplasmic and mitochondrial subtypes respectively and collectively referred to as IDH mutations [IDHm]), and 2) codeletion of chromosome arms 1p and 19q, as observed in oligodendrogliomas. IDH status is typically determined by immunohistochemistry using the anti-IDH antibody that recognizes the R132H-mutated protein (the most common IDH mutation) and/or Next Generation Sequencing. The status of 1p19q is established with fluorescence in situ hybridization. For a description of IDHm see Fig 6.

Fig 6.Click to view

Fig 6.

IDHm. Normally, the IDHwt catalyzes the conversion of isocitrate to α-ketoglutarate (α-KG), forming part of the Krebs cycle and normal cellular metabolism. When mutated, IDHm transforms α-ketoglutarate into 2-hydroxyglutarate (2HG). Because 2-hydroxyglutarate is generated under neoplastic conditions, it is termed an oncometabolite. Accumulation of 2-hydroxyglutarate inhibits α-ketoglutarate-dependent dioxygenases, resulting in hypermethylation of the CpG island in the genome-wide DNA promoter. This increase in methylation inhibits the expression of multiple genes156 but also increases the expression of specific genes like platelet-derived growth factor receptor α (PDGFRα), an oncogene known to drive tumorigenesis.157 This new methylation profile defines a subgroup of gliomas that have a so-called “glioma-CpG island methylator phenotype (G-CIMP).”158 G-CIMP status also silences the expression of various cellular differentiation factors, rendering the tumor cells in an undifferentiated state.159 The G-CIMP low DNA methylation pattern has been associated with shorter survival in IDHm astrocytoma, but additional cohorts are needed for validation to more precisely define the G-CIMP low-methylation diagnostic profile as well as to assess the practicality of testing modalities.48 Me, indicates methylation.

 

The effect of the 2016 WHO Classification System was to generate 5 broad categories of diffuse gliomas: 1) IDHm 1p19q-codeleted oligodendroglial type (further subdivided into diffuse [WHO grade 2] or anaplastic [WHO grade 3]); 2) IDHm non-1p19q-codeleted astrocytic-type; 3) IDHwt astrocytic type (both 2 and 3) further subdivided into diffuse (WHO grade 2) or anaplastic (WHO grade 3); 4) glioblastoma IDHm; and 5) glioblastoma IDHwt. This categorization provided improved prognostic information compared with its predecessors, with IDHm 1p19-codeleted WHO 2 oligodendroglioma type conferring the best prognosis, and IDHwt glioblastoma bestowing the worst (Table 1).

Table 1:Click to view

Table 1:

2016 WHO Classification System of diffuse glioma^a

 

Many of the radiologic clues to the underlying cellular type of diffuse glioma used before the 2016 WHO Classification are still viable (Fig 5). However, because the 2016 classification revealed a clear prognostic benefit for IDHm rather than IDHwt, there was a growing desire to discover radiologic biomarkers for the presence of IDHm. The T2-FLAIR mismatch sign, in which the tumor shows central suppression on FLAIR but not on T2-weighted imaging,50 has high correlation with IDHm, 1p19q-noncodeleted astrocytoma (Fig 7). MR spectroscopy can identify the existence of the oncometabolite 2-hydroxyglutarate in gliomas with IDHm but not in those of IDHwt;51 however, the technique is not used in routine clinical practice because it demands strict processing techniques that limit its reproducibility.

Fig 7.Click to view

Fig 7.

The T2-FLAIR mismatch sign is considered highly specific for IDHm, 1p19q non-codeleted astrocytic gliomas. On T2-weighted images (left), these tumors have areas of fairly homogeneous and strikingly high signal, whereas on FLAIR (right), the corresponding region shows signal suppression with rather a peripheral rim of signal hyperintensity. This appearance is typical of the entity previously known as protoplasmic astrocytoma, which is no longer recognized as a distinct entity in the WHO 2016 Classification of CNS tumors.

 

IDHwt GBMs arise de novo and have been termed primary GBM. IDHm GBMs arise from the malignant transformation of lower-grade astrocytomas and are consequently termed secondary GBMs. Primary GBMs are far more prevalent than secondary GBMs (∼9:1). Clinicoradiologic differences between IDHwt and IDHm GBMs are summarized in Fig 8.

Fig 8.Click to view

Fig 8.

Clinicoradiologic differences between IDHwt and IDHm glioblastoma. Compared with IDHwt GBM, IDHm GBM tends to occur in young patients, has a predilection for frontal lobe location, and demonstrates less necrosis and more unenhancing regions.3,47 IDHwt GBM arises de novo and has been termed primary GBM. IDHm GBM arises from the malignant transformation of lower-grade astrocytomas and is consequently termed secondary GBM. Primary GBMs are far more prevalent than secondary GBMs (∼9:1).

 

The 2016 WHO Classification also contained information relating to molecular aberrations that were commonly associated with the 5 broad types of diffuse gliomas, for example, the presence of amplification of the epidermal growth factor receptor (EGFR amplification) in IDHwt tumors or the homozygous deletion of CDKN2A/2B, especially in IDHm astrocytic gliomas (Table 1). Further insight into how these molecular changes impact diffuse glioma was to lead to updates to the 2016 WHO Classification System.48 It is first necessary to review 3 key molecular pathways that underline diffuse glioma to fully appreciate these revisions.

Molecular Pathways of Diffuse Glioma

Much of the molecular knowledge of diffuse gliomas comes from research by The Cancer Genome Atlas (TCGA) Research Network.52 This work revealed that accumulated mutations, genetic alterations, and epigenetic modifications in key cellular pathways lead to the dysregulation of the cell cycle and permit uncontrollable tumor growth.

The cell cycle is the process in which a cell replicates to generate 2 daughter cells. It is divided into 4 phases: 1) gap 1 (G₁), in which the cell contents (excluding chromosomes) are duplicated and the cell increases in size, 2) S–DNA “synthesis,” whereby 46 chromosomes are duplicated, 3) G₂, further cell growth with organelles and proteins preparing for cell division, and 4) M–nuclear division (mitosis) followed by cell division (cytokinesis). Progression through the cell cycle is controlled at various checkpoints. The G₁ checkpoint, located at the G₁/S transition, is the main decision point for the cell. Passage through this point represents an irreversible commitment to division. Here, certain conditions will be considered to ensure an accurate outcome: for example, assessing that the cell size, level of energy reserves, and the molecular signals are appropriate and also examining for DNA integrity. If conditions are unfavorable, cells may enter a resting phase called G₀ and may re-enter the cycle should the circumstances become satisfactory (Fig 9).

Fig 9.Click to view

Fig 9.

The cell cycle. A 4-stage process (G₁, S, G₂, and M) leading to the generation of 2 daughter cells through cell duplication and separation. After completing the cycle, cells may repeat the process or exit through G₀. From G₀, the cell can undergo terminal differentiation, eg, into astrocytes and oligodendrocytes. Checkpoints are important steps in the cell cycle to ensure the generation of healthy cells.

 

Cyclins are important cell cycle regulators. These proteins activate the enzymes cyclin-dependent kinases (CDKs) and the subsequent complex of cyclin-CDK (C-CDK) phosphorylate target proteins to drive proliferation (eg, G₁/S cyclins activate S-phase targets to allow DNA replication). Three principle cellular pathways have influence over C-CDK activity: 1) receptor tyrosine kinase (RTK), 2) p53, and 3) retinoblastoma (RB). Seventy-four percent of patients with GBM have genetic alterations in all 3 of these pathways.53 Mutations encountered in GBM may be the result of alterations in one or several of the pathways.

RB Pathway.

The retinoblastoma protein (pRB) suppresses cell cycle entry and progression via its interaction with the transcription factor E2F.54,55 Ordinarily, E2F binds target promoters and recruits additional activating proteins to stimulate the expression of cell cycle genes during G₁/S. The binding of pRB to E2F prevents the recruitment of transcription-activating factors and suppresses proliferation.54 The suppressive action of pRB can be overcome by C-CDKs, 55 which, in turn, can be inhibited by p16^INK4a.56 Thus, drivers of proliferation in this pathway are C-CDKs and E2F, and those suppressing cell cycle progression are pRB and the cyclin-dependent kinase inhibitor p16^INK4a (Fig 10).

Fig 10.Click to view

Fig 10.

Key molecular pathways commonly involved in gliomagenesis, especially GBM. 1) The RB pathway. pRB inhibits cell cycle division through its inhibitory action on transcription factors E2F, which ordinarily would promote proliferation. The inhibitory action of pRB can be overcome by cyclin-cyclin-dependent kinase (C-CdK) activity. 2) The p53 pathway. Progression through the cell cycle can be halted by p53 to allow DNA repair or to promote apoptosis. This principally occurs at the G₁/S checkpoint. Through activation of transcription factors such as p21 and p53, C-Cdk activity is suppressed. 3) RTK pathway. The binding of a ligand (eg, EGF) to its RTK (eg, EGFR) can trigger downstream activation of Ras/Raf/MAPK/ Erk leading to promotion of C-CdK activity or PI3K/AKT/MDM2, which suppresses the action of p53. Both pathways act to promote the cell cycle and drive proliferation. Inhibitors of Ras/Raf/MAPK/ Erk include NF1 and p16^INK4a. The action of PI3K/AKT/MDM2 can be inhibited by PTEN (via p53) or by p14^ARF. Both the Ras/Raf/MAPK/ Erk and PI3K/AKT/MDM2 pathways can promote angiogenesis via activation of hypoxia-inducible factor 1α (HIF1α), which subsequently triggers transcription of VEGF. p53, RB, PTEN, NF1, p14^ARF, and p16^INK4a are cell-cycle suppressors, whereas C-Cdks, Ras/Raf/MAPK/ Erk and PI3K/AKT/MDM2 promote cell proliferation. Aberrations in these components can drive tumor growth. mTORC1 indicates mammalian target of rapamycin complex 1.

 

Alterations in this pathway (eg, RB, p16^INK4a, CDK4) result in increased mitotic activity, leading to malignant transformation and have been found in 78% of GBMs.55p16^INK4a is encoded by the CDKN2A gene found on chromosome 9p21. Via a different promoter, the CDKN2A gene also encodes another tumor suppressor protein, p14^ARF,57 which interacts in the p53 pathway.

p53 Pathway.

The gene TP53, located on the short arm of chromosome 17, encodes the tumor-suppressor p53. This protein, also known as “the guardian of the genome” pauses the cell cycle in response to DNA damage and can initiate apoptosis. p53 activates p21, which inhibits CDK4/cyclin D and CDK2/cyclin E complexes and prevents cell cycle progression.55p53 is naturally degraded by the mouse double minute 2 homolog (MDM2), which, in turn, can be inhibited by p14^ARF.2,55 Hence, p53, p21, and p14^ARF act to suppress proliferation while MDM2 promotes it (Fig 10).

Eighty-seven percent of GBM cases studied by the Cancer Genome Atlas Research Network had alterations of the p53 signaling pathway, with p53 being mutated or deleted in 28%–35% of cases.52,55,58 Mutations in TP53 are also found in astrocytomas of multiple grades and may allow early astrocytomas to evade apoptosis.59 Accordingly, p53 alterations are more common in secondary GBMs than in primary GBM tumors.60

Receptor Tyrosine Kinase Pathways.

RTKs are a family of cell surface receptors that bind ligands such as hormones, cytokines, and growth factors. On activation, RTKs signal through 2 major downstream pathways: 1) rat sarcoma virus (Ras)/MAPK/Erk; and 2) PI3K/AK strain transforming (AKT)/mammalian target of rapamycin (mTOR).61 Ras/MAPK/Erk promotes C-CDK activity and drives cell proliferation. This pathway is often activated in certain tumors by mutations in cytokine receptors such as FMS-like tyrosine kinase 3 (FLT-3) or a receptor tyrosine kinase (Kit) or by overexpression of wild type or mutated receptors.62 The tumor-suppressor neurofibromin 1 (NF1) can modify this pathway by inhibiting Ras (Fig 10).63

Binding of a growth factor, such as epidermal growth factor (EGF), to its receptor (EGFR) activates the PI3K/AKT/mTOR pathway and promotes proliferation and survival. AKT leads to the inhibition of several proapoptotic proteins64–66 and the transcription of prosurvival genes67 and modulates MDM2.68 AKT can also activate mTOR, which, in turn, can promote cell growth,55 cytoskeletal organization,69 and induction of hypoxia-inducible factor 1α, the latter leading to downstream activation of VEGF secretion and increased angiogenesis.70PTEN acts as a tumor suppressor by counteracting PI3K signaling. Hence, RTKs and their ligands, Ras (a signaling protein), PI3K, AKT, and MDM2 are drivers of proliferation, whereas NF1 and PTEN act as suppressors (Fig 10).

The most frequent genetic alteration in GBMs is an amplification of the EGFR gene and/or its overexpression at the protein level, which is present in 40%–60% of all GBM cases.71,72 Half of the amplified cases additionally express a truncated mutant variant, EGFR variant III (EGFRvIII). This variant lacks the ligand-binding domain and can be activated in a ligand-independent manner.73EGFR/EGFRvIII expression is associated with increased proliferation and migration of GBMs, contributing to the malignant phenotype of these tumors in an angiogenesis-independent manner.74EGFR amplification has excellent specificity for gliomas with aggressive behavior and is not present in other glioma subtypes that display a more indolent clinical course.75

Mutations in platelet-derived growth factor receptor (PDGFR) may be as high as 60% in GBMs,76 10%–15% in MDM2,77 and mutations in the PTEN gene of 20%–44%.52,58,78,79 Additionally, the signature of combined whole-chromosome 7 gain and whole-chromosome 10 loss (+ 7/− 10) has excellent specificity for gliomas with aggressive behavior.75 Loss of the heterozygosity of the chromosome arm 10q occurs in 60%–90% of GBM cases.80,81 The genes for EFGR and PTEN are located on chromosomes 7 and 10, respectively.

Other Molecular Aberrations Pertinent to the 2016 WHO Classification

Alteration in Telomere-Maintenance Genes.

Following mitosis, the ends of chromosomes (telomeres) undergo a natural process of shortening. Repeat cell division leads to sequential telomere shortening until a threshold is passed and the cell is forced into senescence and/or apoptosis. Most cancers, including gliomas, avoid telomere shortening in 1 of 2 ways: 1) by reactivation of the normally silent enzyme telomerase reverse transcriptase (TERT), or 2) via a collection of TERT-independent mechanisms collectively termed alternative lengthening of telomeres (ALT). ALT is commonly associated with mutations in the α thalassemia/mental retardation syndrome X-linked (ATRX) gene, a central component of the chromatin remodeling complex required for the incorporation of H3.3 histone proteins into telomeres (Fig 11).82

Fig 11.Click to view

Fig 11.

Mechanisms used by glioma to avoid telomere shortening. The TERT and ATRX genes are involved in telomere maintenance.160 Telomeres consist of hundreds of nucleotide repeats that are present at the end of all eukaryotic chromosomes as a nucleoprotein complex.161 These gradually shorten following each round of mitosis, eventually leading to aging.162 Cancer cells characteristically acquire the infinite capability to divide by maintaining telomere length through the sustained expression of telomerase or via the ALT mechanism.163 Typically, TERT and ATRX alterations are mutually exclusive, probably because of functional redundancy.164

 

Loss of ATRX protein expression on immunohistochemistry can be used as a surrogate marker of ATRX mutations with high sensitivity and specificity and is near perfectly correlated with the ALT pathway activation.82TERT mutations are common in IDHwt astrocytic-type gliomas, glioblastomas, and IDHm 1p19q-codeleted oligodendrogliomas, whereas ATRX mutations occur more frequently in IDHm astrocytomas (Table 1).

Beyond 2016 to the 2021 WHO Classification

Hitherto, it has been demonstrated that the terms used to classify diffuse gliomas are deeply rooted in history and based on presumed tumor cell lineage and levels of differentiation. More recently, there has been an appreciation that the diffuse astrocytic gliomas of IDHm and IDHwt represent distinct clinical and genetic entities. Still, there is a need for a better intragroup definition to distinguish, with improved clarity, the grade of diffuse glioma that encompass the IDHm and IDHwt tumors. An improved classification may better guide prognostication and treatment stratification.

IDHm Astrocytic Gliomas.

Recall that the CDKN2A/B gene encodes 2 tumor suppressors: p16^INK4a (involved in the RB pathway) and p14^ARF (p53 pathway). Evidence from multiple retrospective studies suggests that homozygous deletion of CDKN2A/B is associated with shorter survival in patients with IDHm astrocytomas and that its presence corresponds to WHO grade 4 clinical behavior.83–85 Consequently, the Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy (cIMPACT-NOW) suggested incorporating the status of CDKN2A/B into the classification and grading of diffuse astrocytic tumors.48 There was also the recommendation that the term “IDHm glioblastoma” be removed to allow clearer distinction of IDHwt GBM and the grave prognosis that it carries. These have now been incorporated into the 2021 WHO Classification of adult-type diffuse gliomas (Fig 12).

Fig 12.Click to view

Fig 12.

The 2021 WHO Classification of adult-type diffuse gliomas principally concerning changes to the classification of IDHm and IDHwt astrocytic gliomas. IDHm astrocytic gliomas are classified into 3 groups based on histologic features and the status of CDKN2A/B. 1) Astrocytoma, IDHm, WHO grade 2: IDHm astrocytomas that lack significant mitotic activity, histologic anaplasia, microvascular proliferation, necrosis, and CDKN2A/B homozygous deletion. 2) Astrocytoma, IDHm, WHO grade 3; IDHm astrocytoma that contains elevated mitotic activity and histologic anaplasia, but lacks microvascular proliferation, necrosis, and CDKN2A/B homozygous deletion. 3) Astrocytoma, IDHm, WHO grade 4: IDHm astrocytomas with microvascular proliferation or necrosis or CDKN2A/B homozygous deletion or any combination of these features. These tumors have been formerly considered as “Glioblastoma, IDH-mutant, WHO grade 4.” However, they are clinically and genetically distinct from glioblastoma IDHwt and are closely related to WHO grade 2 or 3 IDHm astrocytomas. Significant mitotic activity remains the criterion to distinguish WHO grade 3 from grade 2 tumors. Most neuropathologists use a threshold of 2 mitoses within the entire specimen or 1 mitosis in very small biopsies, while large specimens may require more.47 IDHwt astrocytic gliomas: The minimal molecular criteria for identifying an IDHwt diffuse astrocytic glioma which, despite appearing histologically as a WHO grade 2 or 3 neoplasms, would follow an aggressive clinical course more closely resembling an IDHwt glioblastoma are EGFR amplification or combined whole-chromosome 7 gain and whole-chromosome 10 loss (+7/−10) or TERT promoter mutation. The presence of any of these genetic alterations (even in the absence of histologic features of GBM, eg, necrosis or microvascular proliferation) earns the term “glioblastoma, IDHwt, WHO grade 4.”88

 

Recent research has also suggested that prognosis varies among the IDHm astrocytic subgroup depending on the IDH mutation used as the classifier. Tesileanu et al86 showed that non-IDH1-R132H IDH1/2 mutations are associated with increased DNA methylation and improved survival in astrocytomas, compared with IDH1-R132H mutations. Presently however, IDH1-R132H remains the main classifier of IDH mutation.

IDHwt Astrocytic Gliomas.

The cIMPACT update 3 further recommended that diffuse astrocytoma IDHwt tumors should undergo additional characterization: testing for EGFR amplification; combined whole-chromosome 7 gain and whole-chromosome 10 loss (+7/−10); and the presence of the TERT promoter mutation. Regardless of tumor histology, if any of these molecular alterations are found, the patient should be considered to have diffuse astrocytic glioma IDHwt with molecular features of glioblastoma WHO grade 4.87 These mutations are associated with significantly shorter patient survival compared with patients with other WHO grade 2 or 3 gliomas, and patients have outcomes similar to those of patients with IDHwt WHO grade 4 glioblastoma.87 Subsequently, cIMPACT Update 688 recommended that the nomenclature of these tumors could be simplified further—a move that could facilitate entry into clinical trials—so that an IDHwt diffuse astrocytic glioma could be diagnosed as “Glioblastoma, IDH-wildtype, WHO grade 4” if there is microvascular proliferation or necrosis or ≥1 of the 3 genetic alterations (TERT promoter methylation, EGFR gene amplification, +7/–10 chromosome copy number changes). It is acknowledged that while this recommendation simplifies nomenclature and trial entry, it creates a possible situation in which an IDHwt diffuse astrocytic glioma lacks the histologic hallmarks of glioblastoma (ie, microvascular proliferation, and necrosis) but is still classified as a “glioblastoma.”88 These suggestions have been incorporated into the 2021 WHO Classification (Fig 12).

IDHwt GBM.

Presently, no official update has been made to the WHO classification to subclassify IDHwt GBM. Early clustering analyses of microarray data generated by TCGA proposed 4 distinct GBM subtypes: neural, proneural, classical, and mesenchymal.89 However, subsequent research has suggested that the neural subtype arose from contamination by normal neuronal cells, leading to a refinement of the subclassification to three types instead of four.90 GBM with a classical signature represents a more proliferative phenotype, and those with a mesenchymal signature, a more invasive one; both are associated with worse prognosis. Conversely, the proneural signature represents a GBM subtype associated with better prognosis and an enriched in oligodendrocyte signature (Sox2 and Olig2).91 Adding further complexity is the finding that these subtypes can coexist in different regions of the same tumor and can change with time and through therapy.92,93 Single-cell RNA sequencing has indicated that distinct cells in the same tumor recapitulate programs from distinct subtypes.44 This complex inter- and intratumoural heterogeneity of GBM is ultimately responsible for treatment failure, tumor recurrence, and tumor progression, factors that will be considered in the sections that follow.

Maximal safe resection is recommended for all grades of diffuse glioma. Surgical planning typically involves the use of volumetric MR imaging for neuronavigational purposes. fMRI and DTI may also provide important preoperative assessment of eloquent brain regions and white matter tracts.

Patients who have eloquently sited tumors may undergo awake resection with functional mapping to help preserve neurologic function. It is typical to have involvement from other specialists such as neuropsychologists and speech and language therapists before, during, and after awake craniotomy. If a patient has a radiologically enhancing suspected high-grade glioma (HGG) and the multidisciplinary team (MDT) thinks that surgical resection of all enhancing tumor is possible, then consideration should be given to the use of 5-aminolevulinic acid–guided resection as an adjunct to maximize resection at initial surgery.94 Intraoperative sonography can also be considered to help achieve surgical resection of both low-grade glioma (LGG) and HGG.95 Placement of carmustine implants (Gliadel; Link Pharmaceuticals; biodegradable copolymer discs impregnated with the alkylating chemotherapy agent carmustine) into the resection cavity at the time of surgery may offer a survival benefit in patients with suspected GBM and in whom >90% resection is likely.96 However, this approach has not gained widespread application since the impact of adjuvant systemic chemotherapy with a low risk of side effects is supported by a more robust evidence base and has become standard of care.

For HGG, biopsy should be considered if maximal resection is likely to lead to major functional impairment.97 In exceptional circumstances, the need for pathologic confirmation of GBM may be waived, for example, in elderly patients unfit for diagnostic procedures in whom there is no realistic differential diagnosis based on MR imaging and in whom treatment options are limited.98

Patients who have undergone gross total resection of a histologically confirmed WHO grade 2 diffuse glioma may undergo a period of active monitoring through regular clinical and radiologic surveillance. Active monitoring aims to discover tumor recurrence and the progression of tumor grade. When these circumstances arise, consideration is given to re-operation ± adjuvant chemoradiation. In those with subtotal resection or biopsy alone, consideration will be given to upfront radiation therapy and/or chemotherapy or radiologic surveillance. Delayed repeat fMRI may reveal plasticity of eloquent function (commonly speech and language), meaning that repeat surgery in the initially limited functionally eloquent region becomes possible.

Following maximal safe resection (or biopsy), patients with HGG invariably undergo additional therapy. The National Institute for Health and Care Excellence has recently provided updated guidance for the treatment of newly diagnosed diffuse glioma following surgical resection.94 For patients with GBM, this typically involves concomitant temozolomide (TMZ) and radiation (60 Gy in 30 fractions) followed by 6 cycles of adjuvant TMZ: the eponymous Stupp Protocol. TMZ is a DNA-alkylating chemotherapeutic agent that undergoes spontaneous intracellular conversion via hydrolysis to produce monomethyl triazeno imidazole carboxamide. Monomethyl triazeno imidazole carboxamide is thought to act by methylation of DNA in a way that prevents cell division.99 In patients with GBM, the Stupp regime has demonstrated a significant impact on survival compared with postoperative radiation only: a gain of at least 2 months in median overall survival and a 5-fold increase in survival at 5 years.100,101

Central to the efficacy of TMZ is the methylation status of the methyltransferase (MGMT) gene promoter. Located in chromosomal position 10q26, the MGMT gene encodes proteins involved in DNA repair.102 Specifically, 6–0-methylguanine-DNA methyltransferase (MGMT) is an enzyme that can repair damage induced by TMZ. Some diffuse gliomas undergo epigenetic inactivation of the MGMT gene by promotor hypermethylation, which acts to suppress its transcription and increases sensitivity to TMZ (Fig 13).103

Fig 13.Click to view

Fig 13.

Control of MGMT by promoter methylation. A, Normal production of 6-0-methylguanine-DNA MGMT. This enzyme can repair damage caused by DNA-alkylating agents (eg, TMZ) and can subsequently lead to poor treatment response. B, Epigenetic mutation in some gliomas leads to the hypermethylation (orange circles containing Me) of the MGMT gene promotor. This prevents enzyme production and increases the efficacy of TMZ. Me indicates methylation.

 

MGMT promoter methylation occurs more commonly in IDHm gliomas and has been shown to have a strong correlation with TP53 mutation.104 In HGGs, MGMT promoter methylation is a prognostic indicator of improved survival and decreased progression in patients with IDHm glioma 105 and predicts a favorable therapeutic response to TMZ in patients with IDHwt glioma.106

Despite the existence of promoter methylation, some patients might exhibit ample expression of MGMT messenger RNA. The combination of low promoter methylation and high messenger RNA expression of MGMT could eventually lead to the development of TMZ resistance.107 Most neuroscience centers now test for the median percentage of MGMT methylation to act as a potential predictor of developing resistance to TMZ. Various mechanisms of TMZ resistance have been reported and have recently been reviewed by Singh et al.108 Presently, there are no radiologic biomarkers in clinical use that predict the status of MGMT promoter methylation.

The Stupp Protocol may be adapted in elderly patients with GBM. Phase III studies have shown the effectiveness of shorter courses of radiation therapy in patients 65 years of age or older compared with supportive care alone 109 or standard radiation therapy.110 In the same cohort, the addition of TMZ to short-course radiation therapy (40 Gy, administered in 15 daily fractions during 3 weeks) resulted in longer OS than short-course radiation therapy alone (9.3 versus 7.6 months, P < .001). This prolonged survival was evident irrespective of MGMT promoter methylation, though OS was better in the MGMT methylated subgroup (median OS in methylated MGMT = 13.5 months with radiation therapy plus TMZ and 7.7 months with radiation therapy alone; unmethylated MGMT = 10.0 months with radiation therapy plus TMZ and 7.9 months with radiation therapy alone).111

In patients with grade 2 and 3 IDHm 1p19q-codeleted oligodendroglioma following radiation therapy, 6 cycles of adjuvant chemotherapy (procarbazine, lomustine, and vincristine [PCV]) are often used, given the overall survival advantage observed with the addition of PCV to radiation therapy in phase III trials.112–114 In patients with grade 3 IDHm astrocytoma, an overall survival benefit has been observed with the addition of 12 cycles of adjuvant TMZ following radiation therapy.115

Despite multimodal treatment, diffuse glioma remains a terminal diagnosis with median survival varying with the molecular characteristics of the tumor. IDHm 1p19q-codeleted oligodendroglial glioma with TERT mutation confers the best prognosis (median OS, ∼17.5 years), whereas IDHwt, GBM has the worst (median OS, ∼1.2 years with 25% of patients surviving up to 2 years and a reported 5-year survival rate of just >10%) (Table 1).100,116

Limited prognostic data exists for the further subclassifications of diffuse glioma (as outlined in Fig 12) with the exception that patients with astrocytoma, IDHm WHO grade 2 have a median overall survival of >10 years.84,117 IDHm astrocytomas, WHO grade 4 (formerly considered as GBM, IDHm, WHO grade 4) are clinically and genetically distinct from GBM IDHwt and closely related to WHO grade 2 or 3 IDHm astrocytomas.87 Additionally, IDHwt astrocytic tumors carry a prognosis more closely aligned to IDHwt GBM.118,119 Potential molecular diagnostic and prognostic biomarkers of diffuse glioma are shown in Table 2.

Table 2:Click to view

Table 2:

Glioma tumor markers and their current diagnostic and/or prognostic potential

 

The natural history of diffuse glioma is one of progression with tumor proliferation and brain invasion. Rapid tumor growth demands greater vascularity to deliver glucose and oxygen and to allow the removal of waste. Changes in vascularity and perivascular spread of tumor cells are responsible for some of the radiologic characteristics of diffuse gliomas (Fig 14).

Fig 14.Click to view

Fig 14.

Abnormal tumor vascularity and perivascular spread. Rapid tumor growth demands an increased supply of oxygen and nutrients and a requirement for waste removal, ultimately leading to a drive for vasculature. Gliomas achieve these demands using 4 main methods: 1) vessel co-option, whereby tumor cells group around pre-existing vessels, essentially hijacking them (A, upper left insert); 2) neoangiogenesis, giving rise to new vessels principally through activation of VEGF; 3) vasculogenesis, the recruitment and differentiation of endothelial progenitors from bone marrow or via an angiopoietin receptor (Tie2) and stromal cell-derived factor-1 (SDF-1)/CXCR4 pathways (CXCR4 is a chemokine receptor; its ligand is SDF-1); and 4) vascular mimicry, tumor cells form functional vascular networks similar to real vessels. Glioma cells also spread along the abluminal surface of endothelial cells causing loss of the basement membrane and tight junctions and reduced pericyte and astrocytic endfeet coverage (A). The overall result is the generation of aberrant tumoral vasculature devoid of the normal BBB function, which can be depicted on MRI. B–E, A 56-year-old man with a right temporal lobe GBM. Abnormal vasculature can result in leakage of serum into the brain parenchyma leading to peritumoural high T2 signal (B) and, with further permeability, can allow leakage of gadolinium from the capillary lumen, revealing the enhancing tumor periphery (C). The increased vascularity can be shown using DSC perfusion-weighted imaging through the measurement of relative CBV. Measurements of relative CBV correlate directly with microvessel volume irrespective of enhancement or nonenhancement on T1Gd images.154D, Red horseshoe-shaped signal around the enhancing tumor indicates elevated relative CBV. MR spectroscopy of the tumor core can also demonstrate a lactate peak (E, arrow), indicating the hypoxic-acidic tumoral environment and the typical slope of a GBM with elevated Cho (a marker of cell growth and turnover) and reduced NAA (a marker of normal neuronal health and integrity). Tumor expansion into adjacent brain parenchyma may also cause intravascular thrombosis, vessel occlusion, and hemorrhage (F–H). F, Presenting head CT of a 67-year-old woman shows a hematoma in the right parietal lobe. Delayed MRI (6 weeks, pre- [G] and postgadolinium [H] T1-weighted imaging) reveal the typical appearance of a GBM (central necrosis with peripheral enhancement).

 

Even LGGs that have undergone apparent gross total resection (macroscopically and on imaging) recur and inevitably progress to HGG (Fig 15). Typically, radiologic recurrence of GBM occurs 6–7 months after the operation, quickly followed by clinical relapse.120 Ninety percent of GBMs recur in the peritumoral region (adjacent to the pretreatment enhancing disease).121 Brain infiltration can also lead to new proliferative sites of tumor remote from the originally treated neoplasm and beyond the clinical target volume used in radiation therapy (Fig 16).

Fig 15.Click to view

Fig 15.

Low-grade glioma recurrence and eventual progression in a 36-year-old man with a histology-confirmed IDHm, 1p19q-codeleted oligodendroglioma, WHO grade 2. Columns A–F, FLAIR shows 2 adjacent axial slices (upper row at level of the centrum semiovale; lower row at the cerebral cortex). A, Preoperative imaging shows a well-demarcated diffuse glioma in the posterior aspect of the right frontal lobe. T1Gd images (not shown) revealed no tumoral enhancement. B, Within-72-hour postoperative MRI suggests gross total resection. C–G, Annual postoperative surveillance MRI shows steadily increasing volume of recurrent tumor (high signal intensity) around the resection site (black cavity). Lower image in G (axial T1Gd) shows new enhancement within the tumor nodule at the posterior aspect of the resection cavity (arrow), suggesting dedifferentiation to a higher-grade glioma. H, fMRI performed prior to a redo operation for recurrent tumor indicates left-cerebral-dominant speech activation, remote from the tumor (blood oxygen level–dependent signal, orange yellow) at the posterior aspect of the left frontal lobe, also highlighting tumor involvement in the right precentral gyrus on finger tapping (orange-yellow signal overlying high FLAIR signal of the tumor at the posterior aspect of the resection cavity). As a result, the patient underwent awake surgery to avoid a permanent left-sided neurologic deficit from potential tumor resection in this region. A second histology confirmed progression to WHO grade 3 diffuse glioma.

 

Fig 16.Click to view

Fig 16.

Recurrence of GBM beyond the radiation therapy volume. A 66-year-old man with histologically proven IDHwt GBM, WHO grade 4, TERT-mutated, MGMT-methylated, shown preoperatively in the right frontal lobe as a ring-enhancing lesion (preoperative, upper row, T1Gd) with extensive surrounding high FLAIR signal (preoperative, lower row). An operation achieved gross total resection of the enhancing tumor (postoperative, upper row) with some residual high T2 signal surrounding the resection cavity (postoperative, lower row). Preradiotherapy (RT) planning for target delineation (CT upper row; T1Gd MRI lower row) shows 3 colored demarcations: red, gross tumor volume (GTV) incorporating the surgical resection cavity and any residual enhancing disease; salmon, clinical target volume (CTV) includes GTV and the suspected microscopic disease, typically, a 2.5-cm margin around GTV. Because there are currently no precise imaging methods to locate all of the infiltrative glioma, the CTV invariably includes some tumor but also some normal brain. Blue, planning target volume (PTV). CTV + 5 mm allow uncertainties in the planning or delivery of radiotherapy (eg, CT-MRI fusion and patient setup). PTV is a geometric concept designed to ensure that the radiotherapy dose is delivered to the CTV. It is guided by institutional quality assurance. A 1-year post-RT image showed no tumor recurrence. At <2 years' post-RT, new enhancing tumor is seen in the right parietal lobe (T1Gd upper row) with confluent infiltrative high T2 signal from this tumor to the site of the initially resected GBM in the right frontal lobe (lower row). Subsequent second resection showed IDHwt GBM with molecular genetics similar to the first. RT indicates radiotherapy.

 

In the 1930s Hans-Joachim Scherer, a neuropathologist and one of the pioneers in the study of glioma growth patterns, first suggested that glioma cells infiltrate normal brain parenchyma with distinct morphologic patterns.122 Particularly, he described how invading glioma cells associate with “secondary structures” that constitute elements of existing brain tissue. Diffuse glioma, and especially GBM, made invade via more common paths such as perivascular, through the extracellular matrix, along white matter tracts, and via meningeal and subependymal spread. They may also invade via less frequent routes including through the CSF (leading to drop metastasis) and spread outside of the CNS (Fig 17 and 18).

Fig 17.Click to view

Fig 17.

Glioma invasion through the ECM. The mechanisms underpinning GBM invasion involve both biochemical and biophysical processes that regulate cell shape and its movement across the intercellular space.165 A hypothesis of glioma cell invasion suggests the involvement of the following steps:166 1) detachment of glioma cells from their primary tumor using loss of adherent junctions between adjacent cells and cleaving receptors from their normal ECM ligands (eg, CD44 from hyaluronan); 2) creation of new connections with surrounding parenchyma through the increased expression of integrins and ECM ligands (eg, integrin αvβ5 and vitronectin); 3) destruction and remodeling of the ECM through upregulation of matrix metalloproteinases and plasmin; and 4) migration into healthy tissue by altering cell shape and increasing cell motility through actin-myosin manipulation. These interactions may allow identification of infiltrative glioma, for example, by the presence of increased integrins αvβ3/5, matrix metallopeptidases, plasmin, and urokinase-type plasminogen activator, GTPases, and a visible change in tumor cellular shape in tumor models. Forkhead box-transcription factor (FOXM1B, also a marker of glioma stem cells) links ECM invasion to the molecular pathways previously discussed. FOXM1B upregulates matrix metalloproteinases matrix metalloproteinases-2 (MMP-2) and is capable of transforming immortalized human astrocytes into invasive GBM cells via degradation of PTEN and activation of AKT.35 The PI3K/AKT signaling pathway is also inhibited when urokinase-type plasminogen activator is downregulated, suggesting that this pathway regulates urokinase-type plasminogen activator–induced cell migration.35A–C, A 78-year-old man with an IDHwt GBM in the left temporal lobe (A, T2; B, T1 postgadolinium-weighted imaging). C, DTI shows direct glioma invasion into the adjacent brain and particularly into the left optic tracts (shown as a reduced volume of green white matter tracts running from temporal to occipital [arrows], compared with the normal right side).

 

Fig 18.Click to view

Fig 18.

Routes of glioma invasion: meningeal, subependymal, along white matter tracts, through CSF with drop metastasis, and outside the CNS. A–D, A 48-year-old woman with a biopsy-confirmed IDHwt GBM shown preoperatively as an enhancing mass within the trigone of the left lateral ventricle (A). Eighteen months later, new foci of leptomeningeal disease in the left occipital lobe, left temporal lobe, and more confluent around the cervical spinal cord (B–D, arrows). Also note the resulting hydrocephalus (dilated ventricular system, B and C). E–H, A 72-year-old woman with IDHwt GBM within the medial aspect of the right temporal lobe having subependymal spread at diagnosis (F, arrow revealing abnormal enhancement along the temporal horn of the right lateral ventricle). Eight months later following resection and chemoradiotherapy, there is further subependymal disease (G, arrow), a new dural-based metastasis over the left frontal lobe (H), and new enhancing infiltrative tumor within the white matter tracts of the splenium of the corpus callosum (H, arrow). I–K, A 53-year-old man with an extensive diffuse glioma consisting of an enhancing component adjacent to the left lateral ventricle (I) and infiltrative unenhancing tumor along the corticospinal tracts bilaterally, worst on the left (J and K, arrows). L–M, A 71-year-old woman with a right temporal-occipital IDHwt GBM with concomitant drop metastasis into the spinal canal, including leptomeningeal disease (arrow) and a more masslike intrathecal tumor deposit (arrowhead). N–O, A 71-year-old man with IDHwt GBM in the right temporal lobe (N) with subependymal spread (arrow). At presentation, the patient also had a rapidly growing soft-tissue lesion in his right cheek (O, arrows), confirmed as metastatic GBM.

 

Research has demonstrated that brain regions are differentially invaded by GBM cells, suggesting a varying susceptibility of different areas of the brain to GBM progression, with the hippocampus an especially infrequent site of GBM invasion.123 Although invading glioma cells can become intimately related to blood vessels, they rarely invade the lumen,124 a finding concordant with the long-established clinical observation that gliomas very rarely (0.4%–2%) metastasize beyond the brain.125 Theories that attempt to explain the lack of extracranial disease include the following: 1) Glioma cells may be unable to breach the basement membrane and enter the vasculature;126 2) the extraneural tissues may be unable to support glioma growth; 3) the relatively short survival duration of patients with glioma may preclude the finding of extracranial disease;125 and 4) the immune system limits the potential seeding of circulating glioma cells, a possibility highlighted by occurrence of extracranial GBM in immunosuppressed transplant recipients.127

The infiltrative nature of diffuse glioma means that tumor residuum after surgery is inexorable and implies that the term “recurrence” is an oxymoron. However, the term recurrence is used to signify a return to the disease state as evidenced by clinical and radiologic deterioration. For HGGs, radiologic progression is signified by an increased volume of intravenous gadolinium contrast administration (T1Gd) enhancing disease with concomitant volume expansion of abnormal T2 signal hyperintensity. Radiologically suspected cases of progression of diffuse glioma are invariably discussed among the MDT to allow clinical correlation and management stratification. In patients with focally recurrent diffuse glioma, good performance status and a reasonable interval from first-line treatment, the MDT should consider further surgery or radiation therapy. In those patients with more diffuse recurrence and/or in whom further surgery or radiation therapy is not feasible, TMZ, PCV chemotherapy, or single agent lomustine can be contemplated. Studies of the monoclonal antibody inhibitor of VEGF, bevacizumab, have failed to show an overall survival benefit128–131 and can cause a pseudoresponse on MR imaging, referring to a rapid reduction in tumoral enhancement and vasogenic edema following treatment without a significant change in the volume of actual tumor burden. These imaging changes are attributed to alterations in BBB permeability rather than true antiangiogenic effects.132 Current available evidence also does not support the use of cannabinoid agents, immunotherapy, ketogenic diets, metformin, statins, or valganciclovir, 94 though several of these agents are the subject of ongoing research. When one decides on treatment options, consideration should be given to the performance status of the patient, patient preference, time from last treatment, tumor molecular markers, and nature of the last treatment.94 Pseudoprogression, a transient deterioration in MR imaging appearances mimicking true disease progression, most frequently during the first 3 months after radiation therapy with improvement usually within a few weeks or months,132 should be considered in the context of a well patient but with apparent radiologic progression (Fig 19).

Fig 19.Click to view

Fig 19.

Pseudoprogression (PsP). PsP is radiologically defined as a new or enlarging area of contrast enhancement occurring early after the completion of radiotherapy in the absence of true progressive disease.167 PsP is thought to represent a localized tissue reaction with an inflammatory component and associated edema and abnormal vascular permeability.168 PsP occurs in 10%–30% of patients with GBM undergoing the first MRI following radiotherapy and concurrent TMZ (within the first 12 weeks) and in up to 30%–48% of patients who show image progression within 1 month of the end of radiotherapy.169 Reliable, noninvasive discrimination of PsP from progressive disease is difficult but, in the clinic, is commonly conducted through the use of DSC-PWI. Elevated relative CBV, depicted as red-orange, is suggestive of viable, proliferative tumor, whereas PsP typically shows low relative CBV (blue-green). The figure shows MR imaging of a 55-year-old woman with apparent worsening of left parasagittal motor strip IDHwt GBM, MGMT-methylated with an increase in size and vasogenic edema (A and B) compared with the preceding early postradiotherapy study (C). DWI shows largely facilitated diffusion in the perilesional edema (D and E); tiny foci of restricted diffusion (bright on b = 1000, D, and dark on the ADC map, E) are explained by residual hematoma, shown as intrinsic high T1 signal on the pregadoilinium imaging (not included). DSC-PWI shows no elevated CBF at the center or periphery of the tumor, suggestive of pseudoprogression (F, red signal at the posteromedial aspect of the surgical cavity represents normal relative CBV from the cerebral cortex). Patients with a methylated MGMT promoter have a higher probability of pseudoprogression (91.3%) as opposed to those with an unmethylated MGMT promoter (41%); thus, PsP is more frequently observed after combined chemoradiotherapy, rather than radiotherapy alone.170

 

Mechanisms underlying tumor recurrence are complex. As alluded to earlier, the main challenges underlying therapeutic failure of diffuse glioma and, in particular GBM, are rooted in inter- and intratumor heterogeneity,47 with the latter driven by genetic, epigenetic, and microenvironmental cues that influence cellular programs.44 As previously mentioned, GBM heterogeneity is reflected by the coexistence of different transcriptional cellular subtypes (proneural, classical, and mesenchymal). Further heterogeneity exists in the developmental state of GBM cells whereby the tumor hijacks mechanisms of neural development and contains GSCs that drive proliferation, possess tumor-propagating potential, and exhibit preferential resistance to radiation therapy and chemotherapies.133,134 Treatment may promote the selection of cell subpopulations that are resistant to therapy at the moment of recurrence.135 GSCs can also transdifferentiate into other non-neural lineages like endothelial cells and pericytes, indicating that GBM can actually build its own favorable microenvironment (including intercellular networking) in which GSCs survive despite treatment agents.136 The challenges of treating diffuse glioma also include inadequate drug or agent delivery across the BBB, redundant signaling pathways, and an immunosuppressive microenvironment.137

A detailed review of emerging therapeutic options for gliomas is beyond the scope of this article, and readers are, instead, directed to recent reviews by Pearson et al,55 Le Rhun et al,138 and White et al.45 Briefly, there is a drive to develop new therapeutic agents that target the molecular aberrations that are common to the 3 pathways outlined above (ie, RTK, p53, and RB). These remain in the developmental phase and have mainly involved phase II and III trials. Targets have included RTKs (such as EGFR, vascular endothelial growth factor receptor (VEGFR), PDGFR, insulinlike growth factor type 1 receptor, and fibroblast growth factor receptor, PI3K, mTOR, Ras, rapidly accelerated fibrosarcoma (Raf), p53, and Cdk4/6.⁵⁵ Presently, therapy with RTK inhibitors has failed to provide clinical benefit.139 This may be partly explained by the presence of subclonal mosaicism. Patel et al140 showed that cells within the same GBM exhibited different RTKs, even at the single cellular level. Resistance may also be mediated by dynamic regulation of extrachromosomal mutant DNA.141 Additional therapeutic avenues include chimeric antigen receptor-T-cell therapy,142 oncolytic virus therapy,143 immune checkpoint inhibitors,144,145 neurotrophic tropomyosin receptor kinase fusions and BRAF alterations, 146 and tumor-treating fields.147 Furthermore, there is a desire to better define glioma infiltration by using advanced MR imaging sequences and/or PET. This may allow a move away from the rather generic radiation therapy that is currently delivered and mark a move toward precision medicine. Targeting therapy toward the molecular aberrations of a glioma within an individual patient and better defining tumor volume and biologic behavior with radiation therapy would better support patient-specific treatment.

Fundamental to the management of diffuse gliomas is knowledge of the pathophysiology, diagnosis, treatment, and natural history of this disease. It is evident that diffuse gliomas are a heterogeneous group of primary brain neoplasms that are enriched with molecular aberrations, which permit tumor proliferation and invasion. Diffuse gliomas are driven by the plasticity of GSCs and the interaction between GSCs and the TME. This complex interaction and the ability to adapt drives treatment resistance and leaves the diagnosis of diffuse glioma a poor one. Plasticity also makes it difficult to create a robust classification system because the latter is based on a solitary time point at which the tissue is obtained, one that can change before, during, and after treatment. Nevertheless, recent changes incorporated into the 2021 WHO Classification better align with the biologic behavior of the tumor and patient prognosis.

Imaging continues to play a vital role in the management of patients with diffuse glioma, incorporating various stages of the patient pathway including diagnosis, treatment planning, and surveillance. While there are radiologic biomarkers of disease, these are limited. There is a clinical need for radiology to better depict molecular characteristics of diffuse glioma so that there is provision for noninvasive diagnostic and prognostic biomarkers. Furthermore, imaging should play a vital future role in defining the full extent of tumor invasion, marking the edge of glioma-to-normal brain. This could allow more targeted treatment (surgery and radiation therapy) and provide improved insight into treatment response. Artificial intelligence and computer modeling, especially in medical imaging, is an ever-growing field of research and shows great promise as a future complementary clinical tool. Presently however, more research is needed before this translates into clinical practice.

Finally, more research is needed to gain a comprehensive view of the sources of heterogeneity that underlie diffuse glioma. This is a critical goal for neuro-oncology and has broad implications for any future therapeutics.