Dysembryoplastic neuroepithelial tumors (DNTs) belong to the surgically treatable long-term epilepsy–associated group of tumors. Based on cortical specimens provided through epilepsy surgery at Sainte-Anne hospital, three histologic subtypes (simple, complex, and nonspecific) have been described. Electroclinical data, imaging, intralesional recordings (stereo–electroencephalography [EEG]) and histologic correlations have been recently reviewed in order to assess the relationship between the epileptogenic zone (EZ), the tumor, and associated focal cortical dysplasia (FCD), and to determine optimal strategy for curing epilepsy. Based on a large series (78 patients, 50 male, aged 3–54 years, temporal location 73%, nonspecific forms 68%), we found similar electroclinical data in all DNT subtypes, and demonstrated that magnetic resonance imaging (MRI) features allow differentiation of histologic subtypes. Type 1 (cystic/polycystic-like) always corresponded to complex or simple forms, whereas type 2 (nodular-like) and type 3 (dysplastic-like) corresponded to nonspecific forms. It is notable that we demonstrated intrinsic epileptogenicity in all cases, but found that the EZ differed significantly according to MRI subtype, colocalizing with the tumor in type 1 MRI, including perilesional cortex in type 2 MRI, and involving extensive areas in type 3 MRI. The main prognostic factors for favorable outcome (83% of seizure-free patients) were complete tumor and EZ removal, short epilepsy duration, and lack of cortico-subcortical damage. According to these findings, surgical resection may be restricted to the tumor in type 1 MRI but should be more extensive in other MRI subtypes, especially in type 3 MRI. This MRI-based scheme may be helpful for optimal resection in epilepsy due to DNTs. In addition, we emphasize that early surgery is crucial in curing epilepsy.
Dysembryoplastic neuroepithelial tumors (DNTs) are benign tumors that belong to the category of glioneuronal tumors, typically located in the supratentorial cortex (especially the temporal lobe) and usually revealed as intractable partial epilepsy in young subjects. Since their first description from specimens provided by epilepsy surgery (Daumas-Duport et al., 1988), DNTs have been increasingly recognized in patients with epilepsy. Three DNT histologic subtypes have been described: The initial report consisted of the identification of a specific glioneuronal element (SGNE) associated with glial nodules and focal cortical dysplasia (FCD), corresponding to the so called “complex form.” A variant was secondarily reported: the “simple form,” which presented similar features but was composed of SGNE only (Daumas-Duport, 1993); the “nonspecific form” was finally identified on the basis of the same glial and dysplastic components as observed in complex forms but without SGNE (Daumas-Duport et al., 1999). Despite some controversies, recent studies recognize a range of histologic subtypes of DNTs, including “diffuse” forms that correspond to the nonspecific forms described previously (Thom et al., 2011; Bodi et al., 2012). Complex and nonspecific forms are frequently associated with FCD (type 3b according to the International League Against Epilepsy [ILAE] classification) that may play a role in the epileptogenicity of DNTs. However, the relationship between the epileptogenic zone (EZ) and lesional tissue, and especially the role of FCD, when present, remains incompletely understood.
DNTs are found in about 20% of the histologic diagnoses in adult epilepsy surgery centers (Devaux et al., 2008). However, some adult series in which nonspecific forms are not recognized report a lower rate, around 10% or less (Luyken et al., 2003). In addition, most recent reports focus on pediatric series (Nolan et al., 2004; Bilginer et al., 2009) that do not include nonspecific forms. These selection criteria may account for the discrepancy between series. Despite a high rate of seizure-free patients after surgery (70–90%), surgical failure may occur and repeated procedures are not rare. Therefore, 25 years after the first report, identification of prognostic factors and reliable strategies for optimizing surgical results remain highly relevant.
The main clinical features are similar in each histologic subtype, with a male predominance noted in some series (Sharma et al., 2009; Chassoux et al., 2013). Perinatal events and familial epilepsy are not uncommon (Degen et al., 2002). Epilepsy onset in late childhood (median 7–13 years), intractable complex partial seizures concordant with DNT location, and frequent secondary generalization are characteristic. Infantile spasms are reported in pediatric series (Nolan et al., 2004). Other features consist of few neurologic deficits and cognitive impairment (absent or moderate) and relatively frequent mood disorders (Degen et al., 2002; Nolan et al., 2004; Chassoux et al., 2013).
Electroencephalography (EEG) interictal and ictal discharges are predominantly focal or regional and concordant with DNT location, except in nonspecific forms, in which widespread discharges and discordant abnormalities can be observed. Intrinsic epileptogenicity has been demonstrated by intralesional recordings in each histologic subtype (Chassoux et al., 2013). Typical tumor activity is characterized by rapid spikes or polyspikes against a depressed background activity. This activity may be similar to that found in FCD type 2 (Chassoux et al., 2000), but is usually discontinuous except in a few cases with a severe dysplastic component. Of note, it is never seen in simple forms. Intralesional ictal onset demonstrated by stereo-EEG (SEEG) shows either a slow rhythmic onset pattern (mainly found in simple and complex forms), or a low voltage fast activity that characterizes the seizure onset in nonspecific forms. EZ is always colocalized with the tumor in simple and complex forms, but in only one third of nonspecific forms, in which a widespread EZ including the tumor is predominant. Ictal discharges generated from nonspecific mesial temporal DNTs display an early spread to neocortical structures including extratemporal areas. Conversely, in most neocortical DNTs, mesial structures are involved in the epileptogenic network. The presence of FCD mixed with tumor cortex likely account for this type of EZ organization. However, FCD cannot exclusively explain the discordance between EZ and tumor extent. It is noteworthy that FCD may be difficult to assess within areas of high tumoral cell density and misinterpreted in tumor infiltration zones. We emphasize that, in our experience, FCD is never clearly separated from the tumor and appears diffusely mixed with it in some nonspecific forms (dysplastic-like forms).
The histologic polymorphism of DNTs accounts for the variety in imaging presentations, but common characteristics are cortical location, absence of mass effect, even in voluminous tumors, and absence of perilesional edema (Daumas-Duport et al., 1988; Stanescu Cosson et al., 2001). The most typical magnetic resonance imaging (MRI) pattern consists of a pseudocystic or multicystic appearance, strongly hypointense on T1-weighted and hyperintense on T2-weighted images (Campos et al., 2009). Based on histologic and imaging correlation in a large series (78 patients), we recently demonstrated that different histologic DNT subtypes may be recognizable on MRI, and described the main structural MRI types that allow simple and complex histologic forms to be differentiated from nonspecific forms (Chassoux et al., 2012). MRI features are classified as follows: type 1 (cystic/polycystic-like, well-delineated, strongly hypointense on T1), type 2 (nodular-like, heterogeneous signal), or type 3 (dysplastic-like: iso/hyposignal T1, poor delineation, gray-white matter blurring). We found that type 1 MRI always corresponds to simple or complex DNTs, in temporal and extratemporal areas (Fig. 1). Type 2 MRI (predominant in neocortical areas) and type 3 MRI (mainly in the mesial temporal lobe) correspond to nonspecific histologic forms. Hippocampal sclerosis or malformations (15% of DNTs involving the temporal lobe) and FCD are mainly associated with type 3 MRI (Fig. 2). Calcifications are usually found in type 2 MRI and bone deformation in types 1 and 2. Contrast enhancement is relatively rare (12%) and found equally in all MRI subtypes. True cysts are also rare and are found only in types 2 and 3. In this study, a long preoperative MRI follow-up (up to 16 years) did not show any change in tumor size or appearance. Correlations with neurophysiologic data allowed us to demonstrate that EZ differs significantly according to MRI subtype. It colocalizes with the tumor in type 1 MRI, includes the perilesional cortex in type 2 MRI, and involves extensive areas in type 3 MRI.
Surgical series report favorable outcomes in 70 to 90% of cases (Daumas-Duport et al., 1988, 1999; Hennessy et al., 2001; Luyken et al., 2003; Nolan et al., 2004; Devaux et al., 2008; Bilginer et al., 2009; Campos et al., 2009; Sharma et al., 2009; Chang et al., 2010; Chassoux et al., 2012). Of note, most studies include simple or complex forms only, and there are fewer data concerning nonspecific forms. Surgical planning in reported series is based on electroclinical data combined with MRI and mostly includes intraoperative electrocorticography. Preoperative invasive monitoring (depth or subdural electrodes) is also performed in selected cases. Resections consist either of lesionectomy or corticectomy, including amygdalo-hippocampectomy (AH) and anterior temporal lobectomy (ATL). Whether surgical outcome is related to the size of resection or the type of procedure remains controversial.
Complete tumor removal is considered a major prognostic factor in most studies (Luyken et al., 2003; Nolan et al., 2004; Bilginer et al., 2009; Campos et al., 2009; Chassoux et al., 2012). Conversely, incomplete resection is identified as the main cause of surgical failure, seizure-free outcome being obtained by completeness of tumor removal after a second operation (Bilginer et al., 2009). As in some other reports, we found that patients who underwent reinterventions were not rare (12.8% in our series). Surgical failure has also been attributed to the presence of dysplastic cortex adjacent to the tumor, and removing these areas has thus been considered necessary for obtaining a favorable outcome (Bilginer et al., 2009; Chang et al., 2010), but this has been contradicted by others (Hennessy et al., 2001; Nolan et al., 2004). On the other hand, lesionectomy alone in temporal lobe epilepsy has been associated with a less favorable outcome than ATL (Luyken et al., 2003); this may relate to network organization of the EZ and hippocampal involvement in temporal locations (Thom et al., 2011; Chassoux et al., 2012). We add that this condition is mostly found in nonspecific dysplastic-like forms.
Based on experience provided by SEEG, and correlations between imaging, histologic findings, and surgical results, we described an MRI-based scheme for surgery (Chassoux et al., 2012) that allows avoidance of preoperative invasive procedures. According to MRI features, we propose pure lesionectomy in characteristic, well-delineated cystic or polycystic forms, corresponding to the simple and complex histologic subtypes, in which the tumor and EZ are proven to colocalize. In contrast, we suggest including perilesional cortex in type 2 and type 3 MRI forms, which correspond to histologic nonspecific forms. In addition, we emphasize that AH or ATL should be considered for poorly delineated dysplastic-like (nonspecific) forms.
The second major prognostic factor has been related to young age at surgery and short epilepsy duration (Hennessy et al., 2001; Luyken et al., 2003; Nolan et al., 2004; Chassoux et al., 2012). In addition, we found that successful antiepileptic drug discontinuation was also related to early surgery in young subjects. These findings need to be considered in the treatment plan for epilepsy due to DNTs.
Complications and Differential Diagnosis
Despite some reported cases demonstrating tumor growth and recurrence (Nolan et al., 2004), we think that malignant transformation is unlikely and raises the question of misinterpretation with another diagnosis (particularly gangliogliomas) or complications (intratumoral bleeding or infarction) that can occur in voluminous tumors (Daumas Duport & Varlet, 2003). As in other large series (Campos et al., 2009), we found that recurrence is rare and occurs at the site of residual tumor (only one case without further evolution after reoperation and complete resection in our population). These findings strengthen the importance of performing a complete tumor resection whenever possible.
The main differential diagnosis remains gangliogliomas in which malignant evolution is not rare. Imaging features may suggest a ganglioglioma when a true cyst and associated nodular contrast enhancement are seen; this is much more frequently seen than in DNTs (Daumas-Duport C, personal data). Histologically, gangliogliomas are characterized by the presence of voluminous ganglion-like neurons, which on chromogranin immunostaining typically show a strong and homogeneous expression of this marker, and often exhibit signs of cytoplasmic vacuolization, whereas by definition, ganglion-like neurons are absent in DNTs. However, as ganglion-like cells may be present only focally, they may escape histologic scrutiny. Secondly, gangliogliomas typically exhibit granular bodies and perivascular lymphocytic cuffing, often extending within arachnoid spaces, and they contain an intercellular reticulinic frame. In our experience, these latter features are absent in DNTs. CD34, which is a marker of stem cells, has been specifically related to gangliogliomas; however, we have identified CD34 positive cells in 25% of DNTs in our series, and CD34 positivity has also been reported by others in DNTs, especially in diffuse forms (Thom et al., 2011; Bodi et al., 2012). Chromosomal abnormalities have been reported in gangliogliomas but not in DNTs. However, BRAF (V600E) mutations have been identified recently (Chappé et al., 2013), both in gangliogliomas (44%) and DNTs (30%), including their specific and nonspecific forms. Therefore, searching for this mutation and CD34-positive cells is now recommended in DNTs, especially in the nonspecific forms. The differences in the relative frequency of gangliogliomas versus DNTs in different series reflects the problems posed in differential diagnosis according to World Health Organization (WHO) criteria. In our experience, despite the use of the most specific and restrictive criteria possible, the distinction between DNTs and gangliogliomas may still remain problematic.
The second differential diagnosis that merits discussion (mainly in adults) is low-grade diffuse glioma, when the glial component is exclusively made up of isolated tumoral cells. In these “diffuse” variants (corresponding to the nonspecific “dysplastic-like” DNTs), the cortex–white matter interface is usually blurred, thus rendering the diagnosis particularly difficult. However, on T1-weighted MRI sequences, the diffuse variants of DNTs are isointense to normal cortex, whereas diffuse gliomas are hypointense. In addition, IDH1 mutations are frequent in diffuse gliomas, and although IDH mutations have been reported in a few DNTs (Thom et al., 2011), in our own experience, this mutation is absent in DNTs.
The third diagnosis that may be discussed (mainly in children) is FCD when MRI shows a poorly limited lesion in the temporal lobe with blurring of the grey and white matter junction, also corresponding to nonspecific dysplastic-like DNTs. Of note, the initial radiologic diagnosis was FCD in a recent series focusing on diffuse forms of DNTs (Bodi et al., 2012). However, FCD type 2 lesions are mainly located in extratemporal areas and FCD type 1 lesions are usually associated with normal MRI. Moreover, histologic examination allows identification of tumor cells mixed with the FCD when present (now classified as FCD type 3b).
To conclude, data provided by SEEG demonstrate the high intrinsic epileptogenicity of the three DNT subtypes. However, significant differences concerning the extent of the EZ can be observed. EZ colocalizes with the tumor in simple and complex forms, but is often more extensive in nonspecific forms. Systematic analysis of MRI features offers a scheme for determining the extent of the resection when planning surgery and obtaining the best functional results. We emphasize that complete tumor removal without cortical-subcortical damage represents the best prognostic factor. In addition, long epilepsy duration and older patients at surgery decrease the probability of achieving a seizure-free outcome. Early surgery therefore appears crucial to the cure of DNT-related epilepsy. These data should be taken into consideration when diagnosing DNT after a first seizure in a child.
The authors thank Drs Landré, Devaux, and Turak for their contribution in the care of the patients, and Drs Mellerio and Miquel for the illustrations.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.