Ganglioglioma is probably the most common of the neuronal and glioneuronal tumors, with hundreds reported in the literature (1,2,3,4,5), whereas pure gangliocytomas are rare. Anaplastic gangliogliomas are uncommon, approximately 8% of gangliogliomas in one large series (6).

The usual site for gangliogliomas is the temporal lobe, the location accounting for over three-quarters of all cases (1). The frontal lobe produces most of the remaining cases, with the last few percent occurring elsewhere in the CNS, primarily the cerebral hemispheres, but also rarely including the cerebellum (7), brainstem (8), and spinal cord (9). Males are slightly more prone to developing gangliogliomas in some series, making up from two-thirds of cases (1,10) to just less than half (11). Gangliogliomas can be identified in a wide range of patient ages, with most occurring in the second to fifth decades.

Even more consistent than their tendency to localize to the temporal lobes is the tendency of gangliogliomas to be associated with long-standing partial seizures of greater than 1 year, approaching a rate of 100% in one large series (3). Those not associated with epilepsy do not present with any specific signs or symptoms. No known consistent associations exist between ganglion cell tumors and familial tumor syndromes.

The radiologic appearance of gangliogliomas, aside from location, is variable and nonspecific. Neuroimaging of gangliogliomas shows two general patterns of growth, solid and “cyst with mural nodule,” each that represents about half of all cases (Figure 7-1). A small minority appears entirely cystic on imaging, but diminutive mural nodules can be missed between tomographic slices (5,12). The solid examples tend to be larger, but both are well circumscribed with scant to moderate surrounding edema. Calcification, density on computed tomography (CT), T1 and T2 signal intensity on MR, and contrast enhancement are not consistent in pattern or extent in ganglioglioma (5,12). CT may show erosion of the
skull overlying more superficial lesions (12). A large amount of edema surrounding a ganglioglioma can be associated with more aggressive clinical course.

The clinical outcome for the vast majority of gangliogliomas is favorable, as reflected by their WHO grade I status, with a 7.5-year recurrence-free survival rate of more than 90% (3). Anaplastic, WHO grade III gangliogliomas have dismal clinical outcomes, with median overall survivals of 24.7 and 27 months in two large series (6,13). It is unclear whether these anaplastic cases have a different genetic profile, or have other markers of anaplasia that are nonmorphologic. The presence of BRAF V600E mutations in gangliogliomas presents the opportunity to use targeted inhibitors such as vemurafenib to treat inoperable cases (14,15).

Gangliocytomas contain neuronal cells that vary from large polyhedral ganglion cells with large vesicular nuclei and prominent nucleoli to inconspicuous, small granule neurons. In contrast to the neurons of normal cerebral cortex, those of gangliocytomas and gangliogliomas are more variable in size and shape and have no recognizable architectural organization (Figure 7-2). In contrast to those in gangliogliomas, normal cortical pyramidal neurons are arranged into somewhat distinct layers with their triangular cell bodies all oriented in the same direction like arrowheads pointing to the cortical surface. However, in assessing the architecture of biopsy tissue, it is important to recognize that there is no guarantee that sections will be perpendicular to the cortical surface, or areas that do not have isocortical architecture, such as the amygdala, will be excluded. Tumor
neurons can have multiple nuclei in about half of ganglion cell tumors, a finding that is specific but not sensitive (Figure 7-3). Neurofibrillary tangles are occasionally seen in neoplastic ganglion cells in older patients. The cellular density ranges from very high with back-to-back cells to low, approaching that of normal cerebral cortex. A variable perivascular lymphocytic infiltrate is commonly present, as are eosinophilic granular bodies
(EGBs). EGBs are focal, sometimes intracellular, collections of glassy, eosinophilic spheroids of protein and are found typically in slowly growing, low-grade, primary lesions in the CNS (Figure 7-4). Single, spherical, eosinophilic protein droplets are also common.

The fundamental histologic finding in gangliogliomas is a biphasic tumor cell population with neuronal and glial elements, but the forms of these two elements and the patterns in which they combine are truly diverse. The neuronal element reflects the cytologic features of neurons in gangliocytomas and can range from obvious to subtle, depending mostly on the amount and distribution of the cells. Although neuronal proliferation has been reported in in vitro cell culture of gangliogliomas (16), the neuronal population has not been shown to proliferate in vivo. The glial population is the actively dividing component of gangliogliomas and is the part of the tumor that can recur after excision. The glial cells are also more variable and can display features of fibrillar or pilocytic astrocytoma, or even appear oligodendroglial.

The architecture of ganglioglioma is generally sheetlike or vaguely fascicular and streaming (Figure 7-5), but microcystic myxoid areas similar to dysembryoplastic neuroepithelial tumor (DNT) and pilocytic astrocytoma can be seen (Figure 7-6). The background stroma shows areas of desmoplasia with an extensive network of delicate, reticular collagen fibers. Masson trichrome or reticulin staining highlights this feature when subtle. As with gangliocytoma, a perivascular lymphocytic infiltrate is present in a majority of cases. Vascular proliferation, rare mitoses, and degenerative nuclear atypia are permissible in ganglioglioma, but necrosis
is generally absent, except in anaplastic cases. EGBs are commonplace and Rosenthal fibers can sometimes be seen, yet these also appear in other lesions within the differential diagnosis of ganglioglioma. Small “satellite lesions” of tumor may be present around the edges of the main tumor bulk, but the interface between the lesion and surrounding brain should be well delineated and not show significant infiltration.

Tumor neurons stain similarly to their nonneoplastic counterparts, with reactivity for synaptophysin, chromogranin, neurofilament, class III beta tubulin, MAP2, and NeuN reported (2,4). Neoplastic glial cells show immunoreactivity for GFAP. Any significant immunostaining for p53 is unusual in gangliogliomas and when present, is most often noted in the glial component of recurrent tumors with high rates of MIB1 reactivity (>20%) (2). Other studies have not confirmed significant p53 staining in ganglioglioma (4). MIB1 staining is an important supplement to histology in cases where atypical or anaplastic features are suspected; it is discussed in the following section. CD34 staining is common in the neuronal component in gangliogliomas and could be helpful, if positive, to rule out a nonneoplastic population of entrapped neurons (1) The VE1 antibody to BRAF V600E mutant protein can detect the mutation in gangliogliomas. The typical pattern is strong staining in the neuronal component and weak expression in the glial component (Figure 7-7) (17,18).

The current WHO classification allows for two grades, grade I for most gangliogliomas and grade III for those that are histologically malignant (anaplastic ganglioglioma). Around 8% to 10% of all gangliogliomas will be diagnosed as anaplastic (1,6). Anaplastic cases are associated with a higher MIB1 proliferation index (>5%) compared with grade I gangliogliomas, which are generally around 1% to 3% (2). Histologically, the WHO criteria are not very explicitly defined and include increased mitosis, hypercellularity, and nuclear anaplasia, with microvascular proliferation and
necrosis also mentioned (19). These changes are typically only seen in the glial component. Of the few dozen anaplastic reported gangliogliomas, patient survival has been poor, around a median of about 2 years. Some of the reported cases have been shown to exhibit p53 overexpression, loss of ATRX, H3 K27M mutations and hTERT promoter mutations, perhaps suggesting genetic overlap with other entities (6).

Formerly, a WHO grade II designation was recognized for gangliogliomas that showed increased cellularity, prominent vascular proliferation, and glial nuclear anaplasia. Although the grade II designation is no longer in use, the term atypical ganglioglioma can still be used to communicate the unusual appearance to the treating physician and perhaps prompt closer follow-up.

Depending on location, gangliogliomas may have BRAF V600E point mutations in 18% to 60% of cases, with those in the temporal lobe and posterior fossa having the highest rates (Table 7-1) (8,17,20,21). Because brainstem gangliogliomas are frequently inoperable, the targeted inhibitor vemurafenib can be used successfully to treat progressions of those lesions. Other common genetic features in gangliogliomas include gains of chromosomes 7 (∼20%), or 5 (∼15%) and losses on 22q (∼20%) (22).

Because of their innate tendency to infiltrate among native neurons, diffuse gliomas can give the impression of ganglioglioma on histology. In contrast to the high surgical cure rate for gangliogliomas, infiltrating gliomas are incurable with multiple treatment modalities and progress over time. The importance of distinguishing these two processes from one another cannot be overstated. The first defense against mistaking these two entities is a familiarity with the patient’s clinical history, especially imaging findings. A circumscribed, solid, or solid-cystic mass in the mesial temporal lobe of a patient with a long history of epilepsy would strongly argue
against an infiltrating glioma. Contrast enhancement almost rules out a grade II infiltrating glioma. Histologic features that favor ganglioglioma are multinucleate neurons, clustered or disorganized neurons, fascicular architecture, EGBs, and discreet circumscription. Diffuse gliomas do not show a rich background of reticulin fibers. Because higher grade, IDH wild type diffuse gliomas are unlikely to appear like grade I gangliogliomas, assessment of IDH mutation status, ATRX expression, and 1p/19q deletion will prevent any mischaracterizations.

Of the lesions on the differential diagnosis with ganglioglioma, pleomorphic xanthoastrocytoma (PXA) can be the most indistinguishable due to its overlap in age group, symptoms, location, imaging characteristics, histology, and immunohistochemistry. However, the most important difference between them necessitates their diagnostic separation; PXAs are more aggressive. All of the histologic features of ganglioglioma are seen in PXA, including neoplastic cells resembling dysmorphic neurons, and the genetic feature of BRAF V600E mutations is shared also. Immunohistochemistry and special staining are most useful if the lesion in question is a PXA without ganglionic differentiation. The hallmark of PXA is the presence of large, atypical tumor cells, which should not be present in ganglioglioma. Rarely, PXA and ganglioglioma occur in combination, either as separate foci within the same tumor or as dysmorphic ganglion cells within an otherwise typical PXA. Deletions of CDKN2A/B on 9q are much more common in PXAs than gangliogliomas.

Native nonneoplastic ganglion cells can become entrapped within the substance of pilocytic astrocytomas, imitating the histology of ganglioglioma. Add to that the occasional occurrence of pilocytic astrocytomas in the temporal lobe, the association with epilepsy in this location, overlapping imaging characteristics, and a confusing diagnostic picture can develop. Myxoid microcystic change, EGBs, and Rosenthal fibers are seen in both, so they do not necessarily distinguish. KIAA1549-BRAF fusions strongly favor pilocytic astrocytoma, although BRAF mutations can be seen in either.

DNTs also overlap with ganglioglioma in clinical presentation, location, mixed glial, and neuronal components. The presence of prominent, microcystic, myxoid areas and oligodendroglia-like cells favors DNT, whereas the presence of EGBs, reticular collagen, and lymphocytic infiltrates favors ganglioglioma. The distinction is not of high importance clinically and may be a subjective judgment. Indeed, one may even find examples of composite ganglioglioma/DNT (23).

This low-grade tumor with limited growth potential is strongly associated with drug-resistant, complex partial seizures and was first identified in
temporal lobe specimens removed for seizure control (24). Most patients present with a long history of refractory epilepsy beginning in childhood but can have onset well into adulthood. DNTs have a generally even distribution between the sexes (25,26,27,28).

The classic location for DNT is supratentorial within the cortex of the temporal lobe, often mesially situated, but a large minority of them arises in other areas of cortex and ventricles (29,30). Other noncortical examples have been reported in the cerebellum and brainstem (31,32). Because of their unusual location, ventricular examples arising from the septum pellucidum are frequently misdiagnosed as gliomas, potentially leading to unnecessary treatment (30,33).

On neuroimaging, DNT expands the cortex without showing mass effect or edema (Figure 7-8). Multiple small cystic or pseudocystic structures may be present. T1-weighted MRI shows hypo- or isointensity, while T2-weighted images are hyperintense. Indentation and thinning of overlying calvarium, testament to the slow growth of the lesions, is common. Contrast enhancement is unusual but nodular or ringlike when present. Calcifications are common on CT examination (34). Extracortical cases show similar radiologic features, with the obvious exception of location (Figure 7-9). There is no correlation between radiologic features and histologic type (35).

The calling card of DNT is the specific glioneuronal element, an arrangement of axons clustered in columns that run perpendicular to the cortical surface and are flanked by oligodendroglia-like cells and microcystic pools of acidic mucopolysaccharide material (Figure 7-10). The microcysts
occasionally contain “floating neurons,” a unique characteristic of DNT (Figure 7-11). The oligodendroglia-like cells are monomorphic with round regular nuclei, dark closed chromatin, and prominent perinuclear halos. Fibrillar astrocytes dot the background (24). The larger architecture and configuration of DNT is highly variable and has three general patterns.
These patterns do not have any specific clinical correlations and are more important for the sake of recognition and preventing them from being diagnosed as something else.

The morphologic findings of DNT being diverse and diagnostic interpretation of them somewhat subjective, it might not be surprising that different institutions with different customs in diagnosing DNT would find different rates of specific genetic abnormalities in their cases. Indeed, that seems to be what is recorded in the literature. BRAF V600E mutations have been reported at widely varying rates in DNTs, from 0% (21,41) to 51% (42). Taken together, it seems that BRAF mutations are uncommon in DNTs, particularly when centrally reviewed by expert neuropathologists. Alterations in the FGFR1 gene, which include activating mutations, internal tandem duplications of the tyrosine kinase region, and gene fusions, have been identified in a majority of DNTs and not in other similar tumors (41,43). However, there is a recently described series of tumors that morphologically and immunohistochemically overlap with nonspecific DNTs and have BRAF mutations, or FGFR2 or FGFR3 fusions (44).

Because they both may have a monotonous population of cells with round, regular nuclei and perinuclear halos, often with a microcystic myxoid background, an infiltrative distribution of cells and nodular areas of growth, DNT and oligodendroglioma must both be considered when confronted with a lesion with these characteristics. Although this morphologic overlap is substantial, there are several features that differ. DNTs are centered in the cortex and show a profile of nodular cortical expansion when viewed in cross section, whereas oligodendrogliomas generally involve both the white matter and the cortex. Oligodendrogliomas also tend to have a delicate but extensive network of fine capillaries that are usually lacking in DNT. The nuclear details of the oligodendroglia-like cells in DNT are also more innocent appearing, with condensed chromatin and smaller size. Perineuronal satellitosis by tumor cells is seen frequently in oligodendrogliomas but not as much as in DNT. Although there is no set point at which MIB1 staining becomes indicative of oligodendroglioma, a rate below 1% favors DNT.

In ruling out an oligodendroglioma, assessment of molecular features is mandatory. The presence of an IDH mutation in a DNT-like lesion equals diffuse glioma, and oligodendroglioma if it is also 1p/19q deleted. Some would argue that the absence of those features does not rule out
oligodendroglioma in pediatric patients. However, such “oligodendrogliomas” have excellent clinical courses and have the same FGFR1 alterations seen in the vast majority of DNTs (41,45,46).

Distinguishing DNT from pilocytic astrocytoma, which can show monotonous round haloed cells and mucinous microcysts, can sometimes be difficult histologically but is usually less clinically relevant than ruling out an oligodendroglioma. Although BRAF mutations have been described in both, FGFR1 mutations and duplications are much more typical of DNTs.