Long-term epilepsy-associated tumors

Abstract

The term long-term epilepsy associated tumor (LEAT) encompasses lesions identified in patients investigated for long histories (often 2 years or more) of drug-resistant epilepsy. They are generally slowly growing, low grade, cortically based tumors, more often arising in younger age groups and in many cases exhibit neuronal in addition to glial differentiation. Gangliogliomas and dysembryoplastic neuroepithelial tumors predominate in this group. LEATs are further united by cyto-architectural changes that may be present in the adjacent cortex which have some similarities to developmental focal cortical dysplasias (FCD); these are now grouped as FCD type IIIb in the updated International League Against Epilepsy (ILAE) classification. In the majority of cases, surgical treatments are beneficial from both perspectives of managing the seizures and the tumor. However, in a minority, seizures may recur, tumors may show regrowth or recurrence, and rarely undergo anaplastic progression. Predicting and identifying tumors likely to behave less favorably are key objectives of the neuropathologist. With immunohistochemistry and modern molecular pathology, it is becoming increasingly possible to refine diagnostic groups. Despite this, some LEATs remain difficult to classify, particularly tumors with "non-specific" or diffuse growth patterns. Modification of LEAT classification is inevitable with the goal of unifying terminological criteria applied between centers for accurate clinico-pathological-molecular correlative data to emerge. Finally, establishing the epileptogenic components of LEAT, either within the lesion or perilesional cortex, will elucidate the cellular mechanisms of epileptogenesis, which in turn will guide optimal surgical management of these lesions.

Figure 1

Ganglioglioma. A. Aggregates of atypical ganglion cells of varying size typically with neurofilament positivity demonstrated (inset) are diagnostic criteria. B. Clusters of CD34 positivity cells around dysplastic neurones are often identified with scattered intermingled astrocytic cells (inset). C and D. LEAT with features of diffuse ganglion cell tumor in the temporal lobe with cortex on right side comprising diffuse and subcortical nodular aggregates (arrowhead in C) of neuronal islands with atypical neurones (inset in C) within islands but no glial component (Luxol fast blue/cresyl violet preparation). In the same case (D), neurofilament highlights normal orientation of cortical neurones (arrowhead) as well as the single dysmorphic white matter neurones (arrow in main picture and top inset); CD34 was focally expressed around abnormal neurones (bottom inset). E. A mixed glioneuronal tumor with rarefaction and pallor of the white matter beneath a diffusely infiltrated cortex (arrowhead) but without cavitation or cystic change. (F–H) Cortex adjacent to a ganglioglioma. F. H&E shows a disrupted cortical architecture which may mimic a cortical dysplasia. G. NeuN reveals the vestiges of an overrun cortex with residual lamination, including layers II and IV, recognizable. H. CD34 staining reveals scattered multipolar cells, including in layer I. Scale bar A, B = 40 µm; C = 100 µm; D–H = 115 µm. H&E = hematoxylin and eosin; LEAT = long‐term epilepsy associated tumor; NeuN = neuronal nuclear antigen.

Figure 2

Diagram of the relative distribution of glioneuronal long‐term epilepsy associated tumors (LEATs). DNT = dysembryoplastic neuroepithelial tumor; GG = Ganglioglioma; PGNT = papillary glioneuronal tumor; RGNT = Rosette‐forming glioneuronal tumor.

Figure 3

The relative occurrence of degenerative pathological features in long‐term epilepsy associated tumors (LEATs). These common features attest to the slow tumor growth and progressive accumulation of more “ancient” features. The chart is shown as a five‐tier system from: 1 (palest white shade) = never reported; 2 (next lightest shade) =  rare reports; 3 (mid shade) = infrequent feature; 4 (darker shade) =  often (observed in around half of cases approximately); 5 (darkest shade) = observed in majority of cases. DNT = dysembryoplastic neuroepithelial tumor; GG = Ganglioglioma; PA = pilocytic astrocytoma; PGNT = papillary glioneuronal tumor; PXA = pleomorphic xanthoastrocytoma; RGNT = Rosette‐forming glioneuronal tumor.

Figure 4

Dysembryoplastic neuroepithelial tumor (DNT). Simple‐type DNT with a typical glioneuronal element as visualized with (A) H&E, (B) NeuN and (C) GFAP; in this case, IDH1 mutation was present with positivity for mutant protein in oligodendroglial‐like cells (OLCs), while floating neurones (arrow in D) were immunonegative. This tumor occurred in a patient with a 5‐year history of seizures and a temporal lobe lesion on neuroimaging compatible with a DNT. E. Another simple DNT which had 1p 19q LOH, which is a rare finding. F. Complex multinodular DNT with 1p19q LOH and CD34 highlighting the nodular architecture. G–I illustrate a simple DNT in which intracranial recordings were carried out using subdural grids. The cortical sample illustrated in (H) was from an area with rapid firing discharges and showed normal architecture and no neoplasm whereas a more electrically quiet area (illustrated in I) revealed the DNT. J. A DNT intracortical nodule of NeuN‐negative OLCs (arrowheads) in which entrapped layer V neurones are identified with NeuN. In addition, the entrapped “floating” neurones within DNT‐specific elements display layer‐appropriate cortical laminar markers and normal morphology and orientation, as shown here with Map1b (arrowhead) (K). L. Where a complex DNT overruns the superficial cortex and spills into the overlying leptomeninges, NeuN reveals the underlying residual cortical neurones with occasional NeuN‐positive cells within the meningeal tumor extension (arrowhead, Layers I and II shown). Scale bar = A–E 70 µm; F 80 µm; H,L 120 µm; I–K = 75 µm. GFAP = glial fibrillary acidic protein; H&E = hematoxylin and eosin; NeuN = neuronal nuclear antigen.

Figure 5

Papillary glioneuronal tumor (PGNT). A cystic temporal lobe tumor in a 32‐year‐old male which has shown no recurrence 11 years later. A. Shows papillary projections visible on smear with adherent cells and intervening ganglion cells (toludine blue) and similar composition was noted in cytospin preparations of aspirated cyst fluid (Giemsa) (B). C. Histology showed solid areas composed of mixed glial‐papillary structures and intervening neurocytes. In area D, pseudopapillary architecture was visible. E. Ganglion cells admixed with smaller neurocytes were clearly identified in interpapillary regions and (F) extensive synaptophysin labeling was visible. (G) GFAP expression was restricted to the perivascular layer of cells immediately adherent to the papillae and (H) neurofilament stain highlighted the ganglion cells only. Scale bar = 35 µm. GFAP = glial fibrillary acidic protein.

Figure 6

Rosette‐forming glioneuronal tumor (RGNT) of the fourth ventricle. A–C. This was the third recurrence of an RGNT in 23 years. No anaplastic features were noted but there was vascular thrombosis and hemorrhage in this specimen (not shown): Typical neurocytic rosettes are seen with (A) H&E, (B) synaptophysin and (C) GFAP. (D–F) First resection of a cerebellar vermis RGNT: D. Shows an alveolar arrangement of neurocytes around mucin‐filled cystic spaces. E. Hyalinization of vessels is often a striking feature and (F) proliferating knots of capillaries. Scale bar = 45 µm. GFAP = glial fibrillary acidic protein; H&E =  hematoxylin and eosin.

Figure 7

Neurocytic differentiation in oligodendrogliomas. Case 1 (A–D): Tumor composed of nodules of round cells (A,B) in areas resembling both extraventricular neurocytoma and typical oligodendroglioma, including both focal synaptophysin (C) and GFAP (D) positivity. Case 2 (E–H): oligodendroglioma‐like areas (E) and regions with neurocytic morphology and rosettes (F). There was focal expression of MAP2 (G) and GFAP (inset), in addition to neuronal markers, synaptophysin (H) and NeuN (inset). Both cases showed combined 1p/19q LOH and in patient 2, there was an additional IDH1 mutation, but in neither case was there a seizure history. Scale bar A,B,E 45 µm; C,D,F,G,H 30 µm. GFAP = glial fibrillary acidic protein; MAP2 = microtubule‐associated protein 2; NeuN = neuronal nuclear antigen.

Figure 8

Pilocytic astrocytoma (PA). A–B. (Toluidine blue): intraoperative smear preparations of PA showing cells with fine pilocytic (hair‐like) processes and regular round‐oval nuclei; insert in A: a Rosenthal fiber. C–E: Typical biphasic histological pattern consisting of compact areas and loose, hypocellular areas (C–D, H&E); Rosenthal fibers are shown in D and GFAP immunoreactivity in E. Scale bars: A, D 40 µm; B, 20 µm; C, 160 µm; E, 80 µm. GFAP = glial fibrillary acidic protein.

Figure 9

Pleomorphic xanthoastrocytoma (PXA). A–C (H&E): A: superficially located PXA well demarcated from the underlying cerebral cortex. Panels B–C: PXA with cellular pleomorphism, xanthomatous changes and eosinophilic granular bodies (insert in B); arrows in B show perivascular lymphocytic infiltrates. The xanthomatous changes consist of large cells with accumulation of lipid droplets (C). Panel D.(electron microscopy) shows vacuolated cytoplasm in PXA cells. GFAP (E), MAP2 (F)and synaptophysin (G) immunostaining in PXA cells. Scale bars: A–B, 160 µm; C, F 40 µm; D, 5 µm; E, 80 µm; G, 20 µm. GFAP = glial fibrillary acidic protein; MAP2 = microtubule‐associated protein 2.

Figure 10

Angiocentric gliomas (AG). A–B (H&E): AG with characteristic perivascular pseudorosette growth pattern. C. Compact areas with elongated tumor cells. GFAP (D), EMA (E; “dot‐like” staining), CD34 (F, negative in tumor cells), p53 (G) immunostaining in AG; (images A, B, D–G, kindly provided by Dr. A.J. Becker Department of Neuropathology; University of Bonn Medical Centre, Germany). Scale bars: A: 160 µm; B: 40 µm. C–G: 80 µm. EMA = epithelial membrane antigen; GFAP = glial fibrillary acidic protein.

Figure 11

Model for a proposed future modification of the current classification of LEATs. This is based on histological criteria with addition of “diffuse LEAT” and “mixed LEAT” entities, as well as a LEAT “N.O.S.” category for tumors that remain difficult to classify. Regarding the diagnosis of DNT, these are no longer split into simple and complex subtypes, as there is no evidence to confirm clinical relevance (either in terms of tumor progression or seizure outcome) of categorizing DNT into these two groups. Furthermore, the diagnostic criteria for DNT is based on a constellation of two or more histological features and a glioneuronal element is therefore not an essential criterion. In respect to the presence of cortical dysplasia, this is regarded as a “diagnostic aid” but not a histological criterion. Similarly, molecular genetics and immunohistochemistry (other than GFAP and synaptophysin for assessment of glioneuronal differentiation) are diagnostic aids. The relative contribution of tumor components in the “Mixed LEAT” requires further clarity; however, the additional finding of PA‐like regions in DNT should no longer be regarded as a component of “complex” DNT but diagnosed as a mixed tumor type. “DNT like” tumors arising in the periventricular region are distinct from the cortical based DNTs. DNT =  dysembryoplastic neuroepithelial tumor; GFAP = glial fibrillary acidic protein; IHC = immunohistochemistry; LEAT = long‐term epilepsy associated tumor; LOH = loss of heterozygosity; MG = molecular genetics; OLC = oligodendroglial‐like cells; PA = pilocytic astrocytoma; PXA =  pleomorphic xanthoastrocytoma.