The management of epilepsy is an essential clinical issue in many patients with brain tumors. Tumoral epilepsy is often drug resistant and is associated with poor quality of life. Surgery represents a key therapeutic option in the management of patients with refractory tumoral epilepsy, with high rates of postoperative seizure freedom, especially when gross total resection can be performed. The selection of surgical candidates first requires extrapolation of the presumed underlying pathology and its potential for malignant transformation from clinical and imaging data, especially MRI characteristics. These data determine the decision for surgery, as well as its timing and technical aspects in relation to the risk of postoperative deficit. In glioneuronal tumors, where seizures are often drug-resistant and risk of malignant transformation is very low, epilepsy surgery is usually recommended to alleviate disabling seizures and side effects of antiepileptic drugs. However, the risk of postoperative deficit may outweigh potential benefits of surgery in tumors located within eloquent cortex. This issue is particularly relevant for glioneuronal tumors located within the dominant mesial temporal structures in patients in whom seizure control might require additional hippocampectomy, associated with a high risk of memory decline. In contrast, in patients with low-grade gliomas or aggressive brain neoplasms, both the decision to perform surgery and selection of the best surgical approach primarily rely on the oncologic imperative rather than epileptologic considerations. In these patients, the extent of tumor resection correlates with improved survival, progression-free survival, as well as with the chances of postoperative seizure control.
Brain tumors are one of the most common etiologies of symptomatic focal epilepsy, accounting for around 5% of new-onset seizures and >10% of lesional focal epilepsies (Olafsson et al., 2005). However, both incidence of seizures and their clinical management vary greatly across tumor types (van Breemen et al., 2007; Ruda et al., 2010). Aggressive neoplasms, such as high-grade glioma or brain metastases, can be revealed by seizures, although epilepsy is usually not their key feature and these tumors primarily require rapid oncologic management. In contrast, epilepsy often represents the main clinical issue in patients with non–rapidly progressing brain tumors (van Breemen et al., 2007; Ruda et al., 2010). About 75% of patients with low-grade gliomas and 80–100% of those with glioneuronal tumors, including gangliogliomas (GGMs) and dysembryoplastic neuroepithelial tumors (DNTs), develop epilepsy (van Breemen et al., 2007; Ruda et al., 2010), and drug-resistant epilepsy was reported in about 50% of patients with low-grade gliomas and in up to 90–100% of those with DNTs (van Breemen et al., 2007; Ruda et al., 2010). Patients with brain tumors often show lower tolerability for antiepileptic drugs (AEDs) than patients with epilepsy alone (Glantz et al., 2000). Accordingly, occurrence of seizures is associated with poor quality of life in patients with low-grade gliomas (Aaronson et al., 2011).
It is common practice to offer surgical treatment to patients with epileptogenic brain tumors, either with a view to seizure control, or to optimize oncologic prognosis, or both. Several factors determine the decision to undertake surgery, as well as its timing and technical aspects in relation to the risk of postoperative deficit. The pivotal issue is the presumptive underlying pathology of the epileptogenic lesion, and its potential for malignant transformation. Answering this question distinguishes patients in whom surgery is at least partly motivated by oncologic issues, from those in whom seizure control represents the sole or primary objective of surgery.
The tumor usually bears causal relationship to the seizure focus, with recent evidence pointing to the specific role of the peritumoral rim where increased glutamatergic neurotransmission is observed (Buckingham et al., 2011). Accordingly, complete resection of the lesion, sometimes associated with enlarged resection when extension of the epileptogenic zone (EZ) beyond the immediate peritumoral cortex is suspected, is usually required to control seizures (Englot et al., 2011, 2012b). As in other indications of epilepsy surgery (Ryvlin & Rheims, 2008), the presurgical evaluation must weigh the risks of postoperative deficits and the potential benefit of surgery on seizures control, especially in patients at risk of postoperative neuropsychological deficits when the tumor is located within language or memory networks.
Completeness of tumor resection (Smith et al., 2008) as well as timely surgery (Jakola et al., 2012) impacts the overall survival in both aggressive neoplasms and tumors at high risk of malignant transformation such as low-grade gliomas. Therefore, the evolutionary profile of these tumors requires a surgical decision process different from that applied to more benign glioneuronal tumors.
In clinical practice, the selection of surgical candidates among patients with tumoral epilepsy first requires extrapolating the presumed underlying pathology from clinical and imaging data, especially magnetic resonance imaging (MRI) characteristics (Fig. 1). The following four clinical scenarios have to be distinguished: (1) seizures related to a lesion highly suggestive of a glioneuronal tumor; (2) seizures related to a lesion highly suggestive of a low-grade glioma; (3) seizures related to a lesion for which the differential diagnosis between glioneuronal tumor, low-grade glioma, or nontumoral lesion remains inconclusive; and (4) seizures related to a lesion highly suggestive of an aggressive neoplasm, including high-grade glioma. All four situations will be addressed in detail in the following sections.
Surgery for Lesions Highly Suggestive of Glioneuronal Tumors
Patients with seizures related to DNT or GGM are generally considered good candidates for epilepsy surgery, in particular because seizures usually prove drug resistant.
Presurgical differentiation of glioneuronal tumors from other tumor types, and specifically from low-grade gliomas, relies primarily on their MRI features. DNTs are well-delineated tumors with cystic or polycystic appearance and deformation of the overlying skull (Daumas-Duport et al., 1999; Chassoux et al., 2012). GGMs appear as well-delineated tumors with both solid and cystic parts in about 60%, including frequent calcifications. Nodular or punctuate gadolinium enhancement is frequent (Adachi et al., 2011).
Tumor recurrence after initial DNT resection is rare and the few malignant transformations of DNTs reported so far have been observed exclusively after undue brain radiotherapy (Ray et al., 2009). The situation is different for GGMs, for which a 5% rate of malignant transformation has been reported, as well as slow tumor progression or recurrence after subtotal resection in up to 16% (Southwell et al., 2012). In both tumors, however, seizure control remains the primary objective of a surgical resection.
As for brain tumors in general, the main factor associated with postoperative seizure freedom is the completeness of the glioneuronal tumor resection. A recent systematic review reported that 87% of patients who benefited from gross-total tumor resection were seizure free postoperatively, compared with 55% of those in whom subtotal resection was performed, a finding confirmed by multivariate analysis (odds ratio [OR]5.34; 95% confidence interval [CI] 3.61–7.89; Englot et al., 2012b). The seizure-free rate after surgery did not differ between DNTs and GGMs (Englot et al., 2012b). The risk of seizure recurrence following gross-total lesionectomy was greater for glioneuronal tumors located within the temporal lobe. Accordingly, extending the resection beyond gross-total lesionectomy by performing additional hippocampectomy and/or corticectomy, was associated with higher rates of seizure freedom than gross-total lesionectomy alone (Englot et al., 2012b). This finding suggests that in some patients, the EZ extends beyond the immediate peritumoral cortex; for example, the nonspecific form of DNTs is often associated with focal cortical dysplasia (Daumas-Duport et al., 1999).
Several techniques have been proposed to evaluate the extension of the EZ in tumoral epilepsy and to select the optimal surgical procedure. Intraoperative electrocorticography has been evaluated extensively, but its use was not found to be associated with improved outcome according to one retrospective meta-analysis (Englot et al., 2012b). An MRI-based scheme has been proposed to distinguish the nonspecific forms of DNTs from two other subtypes (simple and complex; Chassoux et al., 2012). Patients whose MRI data fulfil the criteria for the nonspecific subtype of DNT might require an extended cortical resection (Chassoux et al., 2012). It has also been suggested that single-photon emission computed tomography (SPECT) might be helpful to detect cortical dysplasia associated with DNTs (Valenti et al., 2002).
The final decision to perform surgery is always weighed against the risks of the surgical procedure. Given the benign nature of glioneuronal tumors, there are situations where the location of the tumor and its associated risk of postoperative sensorimotor, visual, language or memory deficit, overcomes the potential benefit of epilepsy surgery. However, in patients with severe epilepsy-related impairment, a subtotal resection, with lower surgical risk and lesser chances of seizure freedom than after gross-total lesionectomy, might be proposed (Englot et al., 2012b). The same issue applies to the decision to extend the resection beyond the tumor borders, especially in nonspecific DNTs associated with focal cortical dysplasia, which are often located within the mesial temporal lobe (Daumas-Duport et al., 1999). In this situation, the decision made needs to balance risk of postoperative memory deficit, especially with dominant hemisphere surgery, with theoretical gains in seizure control.
Timing of surgery is another important issue. Most patients with glioneuronal tumors are operated on after demonstration of drug resistance. However, better postoperative outcomes were found with shorter epilepsy durations (Englot et al., 2012b), suggesting that surgery should be offered as early as possible in the course of the disease. This view is reinforced by the fact that the majority of patients with glioneuronal tumors will eventually develop refractory epilepsy (Ruda et al., 2010). Beyond its impact on seizures, early surgery might also improve patients' long-term quality of life by decreasing the length of time at risk for epilepsy-related negative outcomes (i.e., seizures related injuries, antiepileptic drugs adverse events, and risk of sudden unexpected death in epilepsy). The 5% risk of malignant transformation with GGMs as well as the 16% risk of slow tumor progression also favors this approach, and could justify earlier surgery in GGMs than in DNTs. Similarly, the greater risk of tumor recurrence after GGM subtotal resection might justify acceptance of higher surgical risks than in DNTs.
Surgery for Lesions Highly Suggestive of Low-Grade Gliomas
As for patients with glioneuronal tumors, epilepsy surgery deserves to be contemplated in patients with drug-resistant seizures related to low-grade gliomas. However, the decision process that results in the decision to perform surgery greatly differs between the two populations. Malignant transformation is the rule in low-grade glial tumors (Cavaliere et al., 2005). In this context, epilepsy surgery cannot be separated from the management of tumor progression.
The efficacy and predictors of epilepsy surgery in patients with low-grade gliomas do not differ from those reported in patients with glioneuronal tumors, with 71% of patients achieving postoperative seizure freedom, with a significantly greater chance of success after gross-total than after subtotal resection (OR 3.41, 95% CI 2.36–4.93), and lack of benefit from the use of intraoperative electrocorticography to guide extension of the resection beyond gross-total lesionectomy (Englot et al., 2011).
The extent of tumor resection is associated with overall survival and progression-free survival in patients with low-grade gliomas. It has been shown that gross-total resection significantly improves patients' survival in comparison with subtotal resection or tumor biopsy (Smith et al., 2008). Although low-grade gliomas are often located in eloquent areas, a safe resection is usually possible thanks to the brain plasticity enabled by their slow growth (Duffau, 2012). Surgery is facilitated by important advances in preoperative and intraoperative workup (MRI tractography, functional MRI, electrical cortical and subcortical stimulation, and intraoperative MRI, which is an increasingly attractive method to improve resection). Nevertheless, surgery cannot cure grade 2 gliomas, and preserving functions in these patients who might survive decades should remain a priority.
Timely surgery is a critical issue in patients with low-grade gliomas. Considering the impact of seizures on quality of life in patients with low-grade gliomas (Aaronson et al., 2011), drug-resistant epilepsy can be considered a legitimate indication to perform surgery in patients in whom watchful waiting was initially proposed. As in glioneuronal tumors, it should be noted that in patients with low-grade gliomas the postoperative seizure outcome correlates with the duration of epilepsy (Englot et al., 2011). However, the main issue in patients with low-grade gliomas is whether early surgery irrespective of seizure control should be proposed, or rather watchful waiting until malignant transformation or the occurrence of drug-resistant disabling seizures. This issue has long been discussed (Cavaliere et al., 2005), but a recent study reported convincing data suggesting that early surgery significantly improves overall survival and progression free survival (Jakola et al., 2012). Therefore, early and extensive tumor resection might now be considered the best therapeutic option in patients with low-grade glioma. Although this conclusion derives primarily from oncologic observations, it might also prove optimal for seizure management.
Seizures Related to a Lesion for Which the Differential Diagnosis between Glioneuronal Tumor, Low-Grade Glioma, or Nontumoral Lesion Remains Inconclusive
Differentiation between benign tumors and low-grade gliomas relies primarily on the MRI characteristics of these lesions. However, specific features are lacking in a significant proportion of cases, especially for tumors located in the mesial temporal region. Accordingly, misclassifications have been repeatedly reported in mesial temporal DNTs (Stanescu Cosson et al., 2001; Campos et al., 2009). Biopsy might help to confirm the pathologic diagnosis, but it can also lead to misdiagnosis and surgical complications (McGirt et al., 2005). Risk of misdiagnosis has particularly been highlighted in DNTs where biopsy often misses the specific glioneuronal component (Daumas-Duport, 1993). Therefore, biopsy is not recommended for the distinction among DNTs, GGM, and low-grade gliomas, and surgical removal of the lesion is preferred whenever the differentiation between these tumor types is at stake.
Functional imaging using positron emission tomography (PET), discussed elsewhere in this supplement, has been proposed as an additional tool for the diagnosis of non–rapidly progressing brain tumors. Several radiolabeled amino acids have been evaluated, including [18F]fluoroethyl-l-tyrosine (Kasper et al., 2011), alpha-[11C]methyl-l-tryptophan (Juhasz et al., 2006), and [11C]methionine PET (MET-PET; Rosenberg et al., 2005; Rheims et al., 2012). However, the two former tracers might be of limited benefit in distinguishing between DNTs and other tumor types. Increased [18F]fluoroethyl-l-tyrosine uptake as well as abnormal uptake of alpha-[11C]methyl-l-tryptophan has been observed in low-grade gliomas, gangliogliomas, as well as in DNTs (Pichler et al., 2010; Kasper et al., 2011). In contrast, it has been suggested that normal MET-PET findings in patients with an epileptogenic non–rapidly progressing brain tumor might be highly suggestive of DNT, whereas a markedly increased tumor methionine uptake makes this diagnosis unlikely. Among a cohort of 63 patients with focal epilepsy related to a non–rapidly progressing brain tumor who underwent MET-PET, the 13 patients who showed normal MET-PET all had DNTs (Rosenberg et al., 2005; Rheims et al., 2012). It has also been shown that the diagnostic accuracy of MET-PET might be enhanced by semiquantitative analysis where optimal thresholding distinguishes DNTs from other tumor types with 90% specificity and 81% sensitivity (Rheims et al., 2012). Accordingly, the correlation between normal methionine uptake and a pathologic diagnosis of DNT might be strong enough to allow deferring the decision of surgical tumor removal when MET-PET is normal both in seizure-free patients and in patients with tumor located within eloquent cortex. In contrast, increased tumor methionine uptake might result in decision to perform surgery both for diagnosis and therapeutic purposes. In these patients, the extension of tumor resection and the evaluation of the surgical risks should be discussed according to the most likely underlying diagnosis. Given the potential final diagnosis of DNT, conservative surgical procedure might, however, be preferred in most patients.
In the future, MRI spectroscopy might also help to differentiate low-grade gliomas from DNT and GGM through the identification of a specific intratumoral oncometabolite, 2-hydroxyglutarate (2-HG), which results from IDH (isocitrate dehydrogenase) mutation (Andronesi et al., 2012). Because most low-grade gliomas harbor an IDH mutation, in contrast to GGM and DNT where this mutation is never found, the identification of a 2-HG peak on spectroscopy would strongly suggest the diagnosis of low-grade glioma versus GGM or DNT.
Seizures Related to a Lesion Highly Suggestive of an Aggressive Neoplasm
In patients with aggressive neoplasms, including high-grade gliomas or brain metastases, seizures can be disabling but are usually not the key issue. Therefore, oncologic prognosis always outweighs the potential impact on seizures when discussing therapeutic options. Regarding surgery, pathologic diagnosis is mandatory, and extended tumor resection, when possible, might improve overall survival (Sanai et al., 2011). It should be noted, however, that although epilepsy is less frequently observed in these patients than in those with benign tumors or low-grade gliomas (van Breemen et al., 2007; Ruda et al., 2010), drug-resistant seizures in patients with high-grade gliomas are associated with poor quality of life (Yavas et al., 2012), especially in the end-of-life phase (Pace et al., 2013). In this context, the potential impact of tumor resection on seizure frequency should not be underappreciated when discussing indication to perform surgery in patients with high-grade gliomas, and palliative resection might be discussed in patients with refractory seizures (Englot et al., 2012a).
Whatever the tumor type, surgery represents a key therapeutic option in the management of patients with tumoral epilepsy. Tumor resection is associated with seizure freedom in a large proportion of patients, especially when gross-total resection can be performed. In these patients, the main objective of a presurgical evaluation is to distinguish glioneuronal tumors from low-grade gliomas and aggressive neoplasms, since this differentiation determines the acceptable risk–benefit ratio of surgical treatment. The final equation should include the most accurate estimation of: (1) life-expectancy, (2) risk to develop refractory epilepsy, (3) impact of seizures on quality of life, and (4) impact of the various surgical strategies on life-expectancy, epilepsy, function outcome, and quality of life.
The authors have no conflict of interest to report. The authors confirm that they have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.