Keywords

ADC:

Apparent Diffusion Coefficient

AED(s):

Antiepileptic Drug(s)

BBB:

Blood Brain Barrier

CNS:

Central Nervous System

Cho:

Choline

CT:

Computed Tomography

DWI:

Diffusion Weighted Imaging

DNT(s):

Dysembryoplastic Neuroepithelial Tumor

ECoG:

Electrocorticography

EEG:

Electroencephalography

FCD:

Focal Cortical Dysplasia

GCT(s):

Ganglion Cell Tumor(s)

GG(s):

Ganglioglioma(s)

GNT(s):

Glioneuronal Tumor(s)

H&E:

Hematoxylin & Eosin

HS:

Hyppocampal sclerosis

LGG(s):

Low-grade glioma(s)

mI:

myo-inositol

MRI:

Magnetic Resonance Imaging

MRS:

Magnetic Resonance Spectroscopy

NAA:

N-acetyl aspartate

PWI:

Perfusion Weighted Imaging

rCBV:

relative Cerebral Blood Volume

VFD(s):

Visual Field Defect(s)

WHO:

World Health Organization

Glioneuronal tumors (GNTs) are a set of tumors of the Central Nervous System (CNS) composed entirely or partially of cells with neuronal differentiation, diagnosed mainly by means of histochemical and immunohistochemical stainings and, more rarely, by electron microscopy. The description applies to several tumors referred to in the Classification of the World Health Organization (WHO) as neuronal and mixed neuronal-glial tumors (Daumas-Duport et al. 1988; Louis et al. 2007a; Allende and Prayson 2009). Some of them arise tipically in the cerebral cortex and represent a common cause of drug-resistant focal epilepsies in children and young adults (Aronica et al. 2001; Benifla et al. 2006; Cataltepe et al. 2005; Consales et al. 2013; Clusmann et al. 2004; Daumas-Duport et al. 1988; Giulioni et al. 2005; Morris et al. 1998; Nolan et al. 2004). Three groups of tumors are of considerable relevance: ganglion cell tumors (GCTs), ganglioglioma (GG) being the typical example, dysembryoplastic neuroepithelial tumors (DNTs) and neurocytic tumors, namely extraventricular neurocytoma. However, this classification scheme is rather arbitrary and provisional. Tumors with composite features, such as GG and DNT may occur and new entities will probably be described. On the other hand, it could sometimes be difficult to distinguish a tumor from a malformation of cortical development (Prayson and Napekoski 2012). From a more general point of view, it should be kept in mind that also tumors not included in the chapter of neuronal and neuronal glial tumors, should be included in the more general classification of ‘long-term epilepsy-associated tumors’. Some of them, such as the pleomorphic xanthoastrocytoma, may show neuronal differentiation, while some others, such as the pilocytic astrocytoma or ‘isomorphic astrocytoma’, are astrocytic tumors. All of them make clear however that, to induce long-term epilepsy, a tumor need not be composed of neuronal cells (Blümcke et al. 2004).

Furthermore, GNTs commonly arise from a cortex housing developmental malformations, that are able to provoke seizures, such as focal cortical dysplasia (FCD) (40–80 % of cases) and, less frequently (up to 25 % of cases), hyppocampal sclerosis (HS) (Aronica et al. 2001; Bilginer et al. 2009; Blümcke and Wiestler 2002; Cataltepe et al. 2005; Chang et al. 2010; Minkin et al. 2008; Nolan et al. 2004; Prayson et al. 2010; Sharma et al. 2009).

Therefore, management of these cases includes dealing with both tumor and epilepsy.

From the neuro-oncological point of view, consistently with the long clinical history, these tumors are generally considered as low-grade gliomas (LGGs). They correspond to grades I or II of the WHO classification and their therapy relies mainly on surgery (Becker et al. 2007; Daumas-Duport et al. 2007). However, tumor progression or transformation into higher grade tumors may occur (Aronica et al. 2001; Nolan et al. 2004; Hall et al. 1986; Hammond et al. 2000; Kim et al. 2003; Luyken et al. 2004).

From the neurological point of view, it is noteworthy that seizures are likely to respond very well to surgical treatment (Aronica et al. 2001; Cataltepe et al. 2005; Luyken et al. 2003; Clusmann et al. 2002; Clusmann et al. 2004).

Despite the favorable seizure outcome, the best surgical strategy has not been fully established yet. Indeed, while some authors regard tumor resection (the so-called lesionectomy) alone as sufficient for complete seizure control (Giulioni et al. 2005; Bourgeois et al. 2006; Iannelli et al. 2000; Montes et al. 1995), others recommend the resection of peritumoral epileptogenic zones (Aronica et al. 2001; Clusmann et al. 2004; Morris et al. 1998; Cossu et al. 2008; Kim et al. 2008; Luyken et al. 2003).

The aim of this chapter is to elaborate on the current concepts concerning the diagnosis and surgical management of patients with medically intractable focal epilepsy related to GNTs.

Epidemiology and Type of Tumors

GNTs account for 0.4–1.3 % of all brain tumors (Becker et al. 2007; Daumas-Duport et al. 2007) and are an increasingly recognized cause of epilepsy in children and young adults (Aronica et al. 2001; Benifla et al. 2006; Cataltepe et al. 2005; Consales et al. 2013; Clusmann et al. 2004; Daumas-Duport et al. 1988; Giulioni et al. 2005; Morris et al. 1998; Nolan et al. 2004) .

Their incidence in the pediatric population is about 8 %, accounting for up to 30 % of long-standing medically intractable epilepsies (Aronica et al. 2001; Daumas-Duport et al. 1988; Johnson et al. 1997; Obeid et al. 2009; Zaghloul and Schramm 2011). Among telencephalic GNTs (box 1), seizures are reported in up to 100 % of DNTs and in 80–90 % of GGs (Chang et al. 2008; Englot et al. 2011; van Breemen et al. 2007; Zaghloul and Schramm 2011) .

Box 1

Neuronal and mixed neuronal-glial tumors (WHO Classification) (Louis et al. 2007b)

(in bold are tumors relevant for epilepsy surgeons)

Dysplastic gangliocytoma of cerebellum (Lhermitte–Duclos)

WHO Grade I

Desmoplastic infantile astrocytoma/ganglioglioma

WHO Grade I

Dysembryoplastic neuroepithelial tumor

WHO Grade I

Gangliocytoma

WHO Grade I

Ganglioglioma

WHO Grade I

Anaplastic ganglioglioma

WHO Grade III

Central neurocytoma

WHO Grade II

Extraventricular neurocytoma

WHO Grade II

Cerebellar liponeurocytoma

WHO Grade I

Papillary glioneuronal tumor

WHO Grade I

Rosette-forming glioneuronal tumor of the fourth ventricle

WHO Grade I

Paraganglioma

WHO Grade I

Mechanisms of Epileptogenesis Associated with GNTs

The pathophysiological mechanisms of epileptogenesis associated with brain tumors are not well understood. Several pathogenetic factors have been advocated to explain the tumor-related epileptogenesis.

Schematically, three types of mechanisms have been hypothesized, namely: (a) intrinsic to the lesion itself, including the expression of various ion channels and the relative proportion of different cell types within the tumor (Aronica et al. 2001; Ventureyra et al. 1986; Rossi et al. 1999); (b) intrinsic to the tumor location, including peritumoral amino acid disturbances, local metabolic imbalance, cerebral edema, pH abnormalities, neuropil changes, neuronal/glial enzyme changes, altered protein expression and immunological activity (Beaumont and Whittle 2000); (c) unique to the patient harboring the disease (Gaggero et al. 2009; Wetjen et al. 2004).

There is significant evidence in the literature supporting the view that GNTs have intrinsic epileptogenic activities due to their neuronal and glial components (Aronica et al. 2001; Blümcke and Wiestler 2002; Daumas-Duport et al. 1988; Prayson et al. 1995; Prayson et al. 1996; Prayson 1999; Zentner et al. 1994). The neurochemical profile of GNTs shows some similarities in expression of various enzymes and receptors to that of neocortical neurons (Aronica et al. 2001; Aronica et al. 2007; de Groot et al. 2010; Lee et al. 2006). Moreover, the relatively low incidence of HS associated with temporal GNTs suggests that these tumors may be the primary source of epilepsy (Beaumont and Whittle 2000; Blümcke and Wiestler 2002).

The association between GNTs and other epileptogenic lesions (as FCD) has been frequently reported, however the specific role of each of them in determining epilepsy is not well understood (Beaumont and Whittle 2000; Louis et al. 2007b; Prayson et al. 2010). This implicates that simple tumor resection in patients affected also by FCD may determine unsatisfactory seizure outcome.

Finally, it is important to consider that the epileptogenic focus is not located within the tumor in about 30–40 % of patients. The presence of a secondary epileptogenic focus must be postulated, particularly in patients affected by temporal tumors (van Breemen et al. 2007).

Electroclinical Features

Focal epilepsy represents the most common clinical feature. Seizures are often the only symptom. Neurological deficits are quite rare, probably because of the slow growth of GNTs. Epilepsy can present at any age, even if most cases are diagnosed during the adolescence. Seizure type depends on tumor location. Status epilepticus is rare. There does not seem to be differences ascribable to tumor type (DNT, GG, or rarer tumors). The response of GNTs-related epilepsy to antiepileptic drugs (AEDs) is often discouraging, with frequent development of drug-resistance. Short duration of epilepsy, partial seizures and lack of secondary generalization are the most important clinical prognostic factors for a successful seizure outcome. There are two differences between children and young adults concerning the characteristics of GNTs-related epilepsy: a) aura is diagnosed more frequently in young adults; b) mean age at seizure onset is lower in children. This can be explained, respectively, by the limited capability of children to describe their symptoms and their lower seizure threshold (Ozlen et al. 2010).

In the non-invasive diagnostic work-up of patients, long-term neurophysiological monitoring, such as Video-EEG, is very useful in recording seizures and identifying the epileptogenic zone.

Concerning EEG features, interictal data usually consist of spikes and/or sharp waves, sometimes intermixed with slow activities. There is also the possibility of a normal EEG. The abnormal features are usually lateralized to the tumor side. This does not guarantee good seizure outcome after tumor resection. In addition, it should be stressed that satisfactory seizure outcome can be achieved after tumor surgery even in patients whose EEG epileptiform activities are not localized to the tumor site (Morris et al. 1998).

When it is not possible to determine the lateralization of seizure focus by means on non-invasive diagnostic investigations, intracranial EEG (Stereo-EEG, depth electrodes, ECoG) may provide useful information. However, intracranial recordings are mainly aimed at mapping eloquent cortex when tumor resection in critical areas is planned. Thus, little is presently known, for example, about ECoG spike discharge patterns in patients with GNTs (Ferrier et al. 2006).

On the other hand, it is commonly acknowledged that patients with GNTs and FCDs may have similar invasive EEG patterns (Chassoux et al. 2000; Chassoux and Daumas-Duport 2013).

Ganglion Cell Tumors (GCTs)

GG and gangliocytoma are neuroepithelial tumors composed of ganglion cells (gangliocytoma) or both ganglion and glial cells (GG). The distinction seems slightly artificial and seldom acquires a deep, practical significance. Thus, these tumors can be lumped together under the designation ‘GCTs’ .

They are the most common tumors determining drug-resistant epilepsy and account for over 40 % of all epileptogenic tumors in most surgical series (Lawson and Duchowny 2004).

GCTs of the cerebral cortex mainly occur in the temporal lobe and, secondarily, in the frontal lobe (Casazza et al. 1989; Otsubo et al. 1990–1991). They can occur throughout the neuraxis (Becker et al. 2007).

Neuroimaging characteristics are variable (Fig. 1a, b, c, d and Fig. 2a, b) as these tumors may present as a mixed lesion (solid and cystic), as a purely solid mass or, less commonly, as a cyst (Zentner et al. 1994; Castillo et al. 1990; Raybaud et al. 2006). Classically, GG has been described as a cystic mass with a mural nodule in approximately 40 % of cases. On non contrast CT, most GGs are hypodense to gray matter, but some can have mixed or high density. Calcifications are relatively common and are reported in about 35–50 % of cases (Castillo 1998). Scalloping of the calvarium may be seen. Sometimes this neoplasm may be undetectable on CT. On MRI , the signal behavior is variable as well, with a spectrum of signal intensity depending on the neuropathological features of the different components. Solid portions are iso- to hypointense on T1-weighted images, with a variable degree of hyperintensity on T2-weighted images.

Fig. 1
figure 1

GG in a 10-year-old girl.a Axial CT scan. b Axial T2-weighted image. c Axial Gd-enhanced T1-weighted image. d Sagittal Gd-enhanced T1-weighted image. CT shows an heterogeneous, prevailingly hypodense, mass lesion in the right temporal pole with gross, shell-like calcifications (a). The tumor has a cortical location and cystic and solid appearance (b). The solid component is mildly hyperintense on T2-weighted image (b) with marked enhancement following gadolinium injection (c, d)

Full size image
Fig. 2
figure 2

GG in a 10-year-old boy.a Coronal T2-weighted image. b Coronal Gd-enhanced T1-weighted image. Lesion involving the right mesial temporal structures, with mild swelling of the parahippocampal region, hyperintense on T2-weighted image (a). Following gadolinium injection the lesion enhances. No cystic components are present

Full size image

Cystic components, hypointense on T1- and hyperintense on T2-weighted images, may or may not show wall enhancement following gadolinium injection (Zhang et al. 2008). Enhancement of the solid portion is highly variable, ranging from no enhancement to intense homogeneity; grossly half of tumors display at least some degree of enhancement. There is usually little associated mass effect or surrounding vasogenic edema. Intratumoral hemorrhage is uncommon. Leptomeningeal or ependymal seeding may rarely occur (Adachi and Yagishita 2008).

Diffusion weighted imaging (DWI) shows increased diffusivity (high signal on ADC maps), whereas perfusion weighted imaging (PWI) demonstrates lower rCBV values compared to normal parenchyma. On magnetic resonance spectroscopy (MRS), most GGs display the features of a LGG, with decreased N-acetyl aspartate (NAA) and mildly elevated Choline (Cho) .

From the neuropathological point of view, the key feature is the presence of neoplastic ganglion cells. They differ from normal ganglion cells—large cells with vesicular nuclei, central and prominent nucleoli, and abundant cytoplasm—because of abnormalities of size, shape, cell processes, irregular accumulation of Nissl substance and cytoplasmic vacuolization. Cellular gigantism and bi- or multinucleation are a common finding (Fig. 3a). These cells do not achieve any degree of architectural uniformity.

Fig. 3
figure 3

GGa H&E, 200x. Gangliogliomas are typically non-infiltrating tumors; several multinucleated cells, which immunohistochemistry can prove to be neurons, are seen. b H&E. Lymphocytes are common

Full size image

Neurofibrillary tangles, granulo-vacuolar degeneration and basic cellular neuronal lesions typical of neurodegenerative diseases may be observed.

Neoplastic glial cells, whose presence is a requisite for the diagnosis of GG, have a variable morphology, which can make the tumor resemble a pilocytic astrocytoma, a diffusely infiltrating astrocytoma or even an oligodendroglial tumor. Melanotic cells may be found (Soffer et al. 1992).

The stroma is rich in reticular fibers, which encircle not only vessels, as in normal brain, but also neoplastic cells .

Vascular proliferation may be seen. The presence of hemosiderin deposits is an evidence of prior hemorrhage and, like contrast enhancement, is related to an impairment of the blood brain barrier (BBB). Intratumoral BBB dysfunction in concert with subsequent accumulation of albumin by neoplastic glial cells may represent an epileptogenic mechanism underlying tumor-associated long-term epilepsy (Schmitz et al. 2013).

Rosenthal fibers, basic astrocytic lesions, typical of fibrillary gliosis, and eosinophilic granular bodies are often observed. Lymphocytic infiltration is typical (Fig. 3b)

Necrosis is rare except in anaplastic cases.

It is interesting to note that temporal lobe GGs express CD34 glycoprotein, which apparently is not expressed in the frontal or parietal lobe tumors and it could be a marker of epileptogenic dysplasia (Kerkhof and Vecht 2013; Blümcke and Wiestler 2002) .

GGs Grading

The third edition of the WHO Classification of CNS Tumors (Kleihues and Cavenee 2000) admitted the existence of atypical GGs, corresponding to grade II. Atypical tumors should have been distinguished from typical ones on the basis of cellularity, nuclear pleomorphism, microvascular proliferation, and proliferation index (> 5 %). Furthermore, necrosis and a proliferation index (> 10 %) should have been the hallmarks of anaplastic tumors. Unluckily, these criteria are unable to reliably predict the clinical behavior (Wolf et al. 1994; Luyken et al. 2004; Selvanathan et al. 2011). The fourth edition of the Classification includes no longer grade II (Louis et al. 2007b).

Although microscopic signs of atypia or anaplasia mostly occur in the glial component, they may occur in neuronal cells, too (Jay et al. 1994). Although anaplastic tumors typically appear de novo, one tenth of them is the result of the transformation of a GG. Anaplastic transformation is more common in pediatric cases and is associated with previous subtotal tumor resection and radiotherapy (Im et al. 2002).

Dysembryoplastic Neuroepithelial Tumor (DNT)

Although they may occur throughout the CNS, DNTs are cortical-based tumors, that typically affect children and young adults with long-standing pharmacoresistant epilepsy (Daumas-Duport et al. 1988, 2007) .

DNTs of the cerebral cortex typically appear in the isocortex or in the allocortex of the temporal lobe; these tumors can arise in the frontal lobe and, sporadically, in the parietal and occipital lobes (Daumas-Duport et al. 1988; Thom et al. 2011). Although there seems to be a relationship between DNTs and the development of the cerebral cortex, and cortical malformations may coexist alongside, from the neuro-oncologic point of view DNTs are considered as LGGs corresponding to grade I of the WHO Classification (Thom et al. 2011; Louis et al. 2007b).

Initially, the cells of the subpial granular layer were considered the source of this tumor(Daumas-Duport et al. 1988).

DNT coexistence with FCD (Daumas-Duport et al. 2007), neuronal heterotopia (Honavar et al. 1999), microdysgenesis (Rojiani et al. 1996) and neurofibromatosis type 1 (Lellouch-Tubiana et al. 1995) supports the hypothesis of a developmental origin.

DNTs show minimal or no mass effect and absence of peritumoral edema. On CT scan, they appear as a hypoattenuating lesion that may occasionally present areas of calcification and remodeling of the adjacent inner table of the skull. On MRI , the typical appearance consists of a well demarcated pseudocystic lesion, strongly hyperintense on T2- and hypointense on T1-weighted images, with variable FLAIR signal (hypo-isointense or hyperintense). DNTs may have a gyriform or a triangular-shaped pattern with the base pointing to the cortical surface. Hyperintense stripes on FLAIR images are visible both along the surface (bright rim) and on thin septa, resulting in a multicystic, bubbly appearance. Additional small cysts are often located in the vicinity, separated from the main mass.

Some lesions may show a more heterogeneous signal consistent with solid, cystic or semiliquid structures. Solid tissue is usually interspersed between the pseudocysts and it is often found in the adjacent subcortical white matter. Contrast enhancement is rare, variable, and often ring-like (Fig. 4a, b, c, d and Fig. 5a, b, c). Hemorrhage is also uncommon (Daumas-Duport et al. 1988; Daumas-Duport 1993; Ostertun et al. 1996; Campos et al. 2009). Diffusion and perfusion weighted images show respectively increased diffusivity and low rCBV values. The MRS pattern is nonspecific with increase in myo-inositol (mI) and slight reduction of NAA. Lactate and lipids are usually absent (Bulakbasi et al. 2007).

Fig. 4
figure 4

DNT in a 6-year-old boy.a Coronal CT scan. b Coronal T2-weighted image. c Coronal FLAIR image. d Coronal Gd-enhanced T1-weighted image. Right paracentral lobule, cortical-subcortical, lesion whose density (a) and signal intensity on T2 (b) is cystlike. Corresponding FLAIR image (c) shows bright internal septa and lesion periphery producing a multicystic “bubbly” appearance. Following gadolinium administration enhancement is absent (d)

Full size image
Fig. 5
figure 5

DNT in a 9-year-old girl.a Axial T2-weighted image. b Axial FLAIR image. c Axial Gd-enhanced T1-weighted image.Superficially located tumor involving the right postcentral gyrus with remodeling of the adjacent calvarium (a). Two small internal pseudocysts are visibile on FLAIR image (b). There is no enhancement on post contrast T1-weighted image (c)

Full size image

The neuropathological key feature is a microscopic structure, the specific glioneuronal element, which assumes the leading role in the simple forms, sharing the stage with the glial nodules in the complex forms. A highly controversial set of neuroepithelial tumors lacking the specific glioneuronal element has been classified as nonspecific or diffuse forms (Thom et al. 2011; Honavar et al. 1999; Bodi et al. 2012).

Simple and complex DNTs are generally circumscribed cystic lesions corresponding to type 1 of some neuroimaging classifications. The diffuse forms of the neocortex are quite nodular and sometimes calcific (type 2), the diffuse forms of the allocortex (mesial temporal lobe) tend to have dim outlines (type 3) (Chassoux and Daumas-Duport 2013; Chassoux et al. 2012).

The glioneuronal element is characterized by a mucoid matrix in which cells resembling oligodendrocytes align in a columnar fashion along bundles of axons and capillaries arrayed perpendicular to the pial surface. Mature neurons seem to float among the columns (Fig. 6a, b, c). Binucleate cells can be present; perineuronal satellitosis is not generally seen.

Fig. 6
figure 6

DNTa Low-power of a DNT. b H&E, 200x. Bundles of axons and capillaries cross a mucoid matrix, in which some pyramidal neurons float. c H&E, 200x. Columns of cells resembling oligodendrocytes line up around axons. d H&E, 200x. Some areas may resemble to an pilocytic astrocytoma or ganglioglioma

Full size image

The glial nodules, which may contain neurons, are made up of astrocytes and oligodendrocytes, showing a variable degree of differentiation or pleomorphism. The nodules may resemble pilocytic astrocytoma or ‘diffuse gliomas’ (Fig. 6d).

Calcifications may be observed (Daumas-Duport et al. 1988; Thom et al. 2011); ossification is an exceptional finding (Thom et al. 2011).

These tumors have the capability both to cross the cortical sulci and to extend into the white matter and the leptomeninges (Thom et al. 2011; Zhang et al. 2013). White matter may show loss of myelin, gliosis and mycrocysts (Thom et al. 2011). HS, more commonly atypical (end-folium sclerosis) than classic, is common (Thom et al. 2011). Neurofibrillary tangles may be observed. The presence of intracellular pigments, such as iron or melanin, is quite common (Thom et al. 2011).

Mitotic figures are very rare in the typical cases and the proliferative index is low. Although prominent vessels can be seen, endothelial proliferation is generally absent.

Necrosis is unexpected.

These tumors are able to regrow even after apparently complete resections showing a typical or atypical morphology (Daghistani et al. 2013; Schittenhelm et al. 2007). In some cases tumors seem to turn into something similar to anaplastic astrocytomas (Ray et al. 2009), glioblastoma (Duggal et al. 2008; Chuang et al. 2013) or oligoastrocytoma (Gonzales et al. 2007).

Neurocytic Tumors (Extraventricular Neurocytoma)

Neurocytic tumors of the cerebral cortex may cause focal drug-resistant epilepsy (Giulioni et al. 2011). These rare tumors, corresponding to grade II of the WHO Classification, typically appear in the frontal lobe, although they can arise in the parietal lobe and, sporadically, in the temporal and occipital lobes (Brat et al. 2001).

Temporal lobe tumors may coexist with FCD (Giulioni et al. 2011) .

Neurocytic tumors are well demarcated lesions with variable and nonspecific appearance on neuroimaging studies depending on cellularity and anatomic location. On plain CT scan, they can appear as hypodense or isodense to gray matter; areas of patchy calcifications are common. On MRI , they have been described as hypointense or isointense on T1- and iso- to hyperintense on T2-weighted images with heterogeneous enhancement following the administration of a contrast agent. They commonly display cystic areas and mild to moderate peritumoral edema; intratumoral hemorrhage can occur. DWI studies showed nonspecific features with variable degree of diffusivity. On MRS, strongly decreased or no NAA peak and prominent Cho peak have been reported (Liu et al. 2013; Tortori-Donati et al. 1999; Han et al. 2013; Patil et al. 2014).

The key neuropathological feature is the presence of neurocytes, which are small round cells with modest quantity of cytoplasm, which can look clear, and nuclei with finely granular chromatin and one or more small nucleoli (Fig. 7a, b). Larger ganglioid cells or even larger ganglion cells may be found (Giangaspero et al. 1997; Tortori-Donati et al. 1999). Tumor cells arrange in sheets, clusters, ribbons, or rosettes, with fine neuropil separating cell aggregates (Scheithauer et al. 2001). Immunohistochemical stainings consistently demonstrate synaptophysin, a major synaptic vesicle protein, in the neurocytes.

Fig. 7
figure 7

Neurocytoma. a H&E, 200x. Neurocytomas are typically composed of uniform cells, whose clear halo may be prominent. b H&E, 100x. Calcifications are common

Full size image

Some tumors (ganglioglioneurocytomas) harbor an astrocytic component.

The mitotic index is generally inconspicuous. Atypical features (high mitotic index, vascular proliferation, necrosis) and high cell proliferation may herald an aggressive clinical behavior.

Multinodular and Vacuolating Neuronal Tumor (MVNT)

This new entity is a purely neuronal tumor of the cerebral hemispheres, which seems to arise mainly from the temporal lobe, displaying a benign clinical behavior. This tumor has been associated with intractable epilepsy (Bodi et al. 2014; Huse et al. 2013).

Surgical Strategies and Complications

Surgery can be considered the treatment of choice for GNTs-related epilepsy. Focal epilepsies caused by GNTs are scarcely responsive to antiepileptic drugs and the surgical treatment of these epileptogenic tumors can offer seizure-freedom in up to 90 % of cases (Daumas-Duport et al. 1988; Giulioni et al. 2005; Kirkpatrick et al. 1993).

Furthermore, surgical resection prevents tumor growth and the risk of anaplastic transformation. Last, the surgical approach leads to neuropathological diagnosis.

Although surgery in patients with GNTs-related epilepsy yields a good seizure outcome, the optimal surgical strategy for these tumors has not been fully established. In fact, some authors consider that tumor resection alone (the so-called “lesionectomy”) is sufficient to achieve complete seizure control (Bourgeois et al. 2006; Giulioni et al. 2005; Iannelli et al. 2000; Montes et al. 1995). On the other hand, other investigators also recommend the additional resection of peritumoral epileptogenic zones to maximize the seizure outcome (Aronica et al. 2001; Clusmann et al. 2004; Cossu et al. 2008; Morris et al. 1998; Kim et al. 2008; Luyken et al. 2003).

In the surgical management of GNTs, it is of paramount importance to consider some differences between these tumors and other groups of low-grade intra-axial neoplasms. First of all, GNTs are benign tumors in which epilepsy is generally the only clinical feature. Conversely, other types of LGG may present many other clinical findings (intracranial hypertension, focal neurological deficits, etc.). Second, unlike other LGG, GNTs are often located on temporal or temporo-mesial regions. Third, the pathophysiological mechanism of epilepsy in GNTs differs greatly compared to other LGG. GNTs have a so-called intrinsic epileptogenicity due to their neuronal and glial components. In addition, other epileptogenic diseases (above all, FCD) may be associated with GNTs, may contribute to determine the epileptogenic zone in some patients and are not always well distinguishable from the tumoral tissue on MRI . Finally, in temporal GNTs, the hippocampus may contribute to the epileptogenesis of these lesions even without apparent involvement at MRI or pathological examination (Chang et al. 2008; Englot et al. 2011; Smits and Duffau 2011; van Breemen et al. 2007; Blümcke et al. 2011; Tassi et al. 2002; Giulioni et al. 2009; Schramm and Aliashkevich 2008; Morioka et al. 2007).

Considering these features, the target of surgery for GNTs to obtain the best seizure outcome is not only the anatomical (tumoral) lesion, but the entire epileptogenic zone, which sometimes is larger than the simple structural lesion. Thus, the surgical strategy must be based on the anatomo-electro-clinical correlations that are necessary to identify the epileptogenic zone (Lüders et al. 2006). This principle is particularly important especially for lesions located in the temporal lobe, where the epileptogenic network is usually much more complex and extended than the tumoral area.

Thus, from the surgical point of view, distinctive “philosophies” of treatment may be adopted, usually depending on the location of the tumors.

For lesions located in extratemporal regions the target of surgical resection is the tumoral lesion only. In other words, only a so-called “lesionectomy” is performed (Bourgeois et al. 2006; Montes et al. 1995).

For lesions located in the temporal lobe (the most frequent site of GNTs), several approaches have been described, including lesionectomy, extended lesionectomy, tailored resection and anterior temporal lobectomy (Casazza et al. 1997; Clusmann et al. 2004; Cossu et al. 2008; Jooma et al. 1995; Quarato et al. 2005).

To our knowledge, presently few studies correlate the best surgical strategy to obtain the best seizure outcome to tumor location (Cataltepe et al. 2005; Giulioni et al. 2009; Luyken et al. 2003). There is some consensus that the best seizure outcome for epileptogenic GNTs located in extratemporal and temporo-lateral regions is provided by simple lesionectomy. For temporo-mesial GNTs, the results of a simple lesionectomy are not particularly encouraging, while more extended surgeries (tailored resection) offer better seizure outcomes (Giulioni et al. 2009).

Finally, another important issue concerning the surgical strategy for GNTs is the additional resection of the hippocampal-parahippocampal complex when it is not invaded by GNTs and does not show other signal abnormalities on MR imaging. In fact, if a more extended surgery than simple lesionectomy for temporo-mesial GNTs is a generally accepted concept (Schramm 2008), the correct balance between the extent of resection necessary to provide the best seizure outcome and the avoidance of neuropsychological deficits is still an open problem deserving further studies (Helmstaedter et al. 2011).

Neurosurgery has become safer in the last decades, thanks to the improvement of diagnostic and operative techniques and anaesthesiological procedures.

However, craniotomy and brain surgery imply the risk of complications, whose rate is quite low; mortality is between 0.5 and 1 % (Pilcher and Rusyniak 1993).

The neurosurgical complications of GNTs surgery can be subdivided into craniotomy-related complications (general neurosurgical complications) and complications more specifically related to resective brain surgery.

The first group is represented by hematomas (extradural, subdural, and intracerebral), infections, cerebrospinal fluid leak and, rarely, air embolism and pulmonary embolism. These complications are responsive to appropriate treatment and do not usually influence surgical outcome.

Concerning the second group of complications, we can distinguish between major complications (when severe neurological deficit or daily activity impairment occur) and minor complications (disappearing within 3 months).

Neurological complications depend on the site of resection. Considering that lesionectomy and tailored resection are the surgical strategies commonly adopted in the management of GNTs and these neoplasms are preferentially located in the temporal lobe, typical major neurological complications of temporal lobe surgery (TLS) include contralateral hemiparesis or hemiplegia, homolateral third nerve palsy or paresis, contralateral visual field defects (VFDs) and speech disturbance (Pilcher and Rusyniak 1993; Polkey 2004).

The hemiplegia or hemiparesis can be related mainly to vascular causes (manipulation of middle cerebral artery branches and/or other arteries) (Polkey 2004).

VFD consists classically in superior quadrantanopia contralateral to the resection (Egan et al. 2000). This finding is caused by injury of Meyer’s loop (the most anterior part of the optic radiation). Other VFDs in temporal lobe surgery can also occur because of injuries of the optic tract or lateral geniculate body. However VFDs should be considered as an expected event in TLS. Today recent advances in neuroimaging techniques as diffusion tensor imaging (DTI) allowing a better anatomic definition of the optic radiation could contribute to reduce post-resection VFDs.

Other rare complications of TLS include hemorrhage distant from the site of surgery (Toczek et al. 1996; Giulioni et al. 2006; Yacubian et al. 1999), probably due to transmural venous pressure variations (Giulioni and Martinoni 2011) and middle fossa cyst causing raised pressure (Weaver et al. 1996). The global incidence of major neurological complications of TLS is 0.37–4 % (Pilcher and Rusyniak 1993; Polkey 2004).

Surgery-related complications in frontal lobe GNTs include hemiparesis if the resection encroaches upon the gyrus anterior to the precentral gyrus. Broca’s area must be preserved in the dominant hemisphere. Some authors recommended such resections should be performed under local anaesthesia (Polkey 2004) and/or using a certain number of localizing techniques such as functional MRI and transcranial magnetic stimulation for motor cortex localization (Macdonell et al. 1999).

In spite of these precautions, any resection centered on the central area (primary motor and sensory area) carries the risk of loss of its function.

Resections for lesions located in parietal and posterior temporal regions of dominant hemisphere can determine receptive aphasia, dyslexia, disgraphia and dyscalculia.

Finally, for occipital resections, some degree of VFD(s) can be expected, even if some authors noted that the patients adapted to their VFD within 1 year after surgery (Williamson et al. 1992).

Seizure Outcome

The post-operative seizure outcome is currently assessed according to Engel’s Classification (Engel 1996).

The rate of seizure-free outcomes (Engel Class I) in GNTs –related focal epilepsies is about 90 %.

Several factors have been correlated with seizure outcome: extension of surgical resection, histotype, duration of epilepsy, age at surgery, association of the tumor with perilesional cortical dysplasia (Aronica et al. 2001; Daumas-Duport et al. 1988; Im et al. 2002; Morris et al. 1998; Nolan et al. 2004).

The main predictor of excellent seizure outcome is the complete removal of the tumor (Morris et al. 1998; Nolan et al. 2004; Khajavi et al. 1995; Kim et al. 2001).

Concerning the completeness of resection, it must be emphasized that only few studies attempted to establish a correlation between tumor site (i.e., temporo-mesial versus temporo-lateral and extra-temporal) and surgical strategy (lesionectomy versus resection of the epileptogenic focus, including the tumor).

Giulioni et al. (2005, 2006) observed that patients affected by extra-temporal and temporo-lateral GNTs had better seizure outcome than those with temporo-mesial GNTs after lesionectomy alone. In addition, in a study comparing retrospectively seizure outcome between two homogeneous series of temporo-mesial epileptogenic GNTs treated respectively by lesionectomy alone and tailored surgery, the same authors reported a much better seizure outcome (93 % vs 42.8 %) in patients who underwent tailored resection (Giulioni et al. 2009).

These observations and other evidences concerning the role of the temporo-polar cortex in temporal lobe seizures (Chabardès et al. 2005) and the frequent association of GNTs with other epileptogenic abnormalities (Daumas-Duport et al. 2007) suggest that tailored resection for temporo-mesial GNTs is to be preferred to obtain a better seizure outcome.

Future Trends

In the future, optimal therapeutic management of GNTs will need a better understanding of the epileptogenic mechanisms related to these neoplasms.

In addition, further studies are required to elucidate GNTs preference for the temporal lobe.

Neuronal precursors in the subgranular zone of the dentate gyrus and the occurrence of postnatal life neurogenesis have been recently described (González-Martínez et al. 2007; Paradisi et al. 2010; Siebzehnrubl and Blümcke 2008). These data could contribute to explain more complex epileptogenic mechanisms and/or the occurrence of GNTs.

Moreover, the well known association between GNTs and FCD needs further investigations. If common precursor cells for these two diseases can be hypothesized, either at an embryological stage or during postnatal life, it still remains to be clarified whether FCD coexists with GNTs or has some potential to change into neoplastic cells (Daumas-Duport et al. 2007).

Conclusions

GNTs are one of the most common causes of drug-resistant epilepsy in children and young adults. GNTs-related epilepsies usually respond very well to surgical treatment with an extremely favorable seizure outcome and important effects on cognitive development and patients’ quality of life.

Surgical treatment should be planned on the basis of anatomo-electro-clinical correlation defining the epileptogenic zone (which may involve a larger area than the tumor) to be resected.

Tumor location is another important factor (temporo-mesial, temporo-lateral, extratemporal site) to be considered in planning surgical resection.

Finally, even in the event of sporadic seizures and/or seizures responsive to AEDs, surgery should be proposed in order to obtain complete seizure control with freedom from AEDs, and to prevent tumor growth and risk of malignant transformation.