Sturge–Weber Syndrome Is Associated with Cortical Dysplasia ILAE Type IIIc and Excessive Hypertrophic Pyramidal Neurons in Brain Resections for Intractable Epilepsy

Abstract

Sturge–Weber syndrome (SWS) is a rare syndrome characterized by capillary‐venous malformations involving skin and brain. Many patients with SWS also suffer from drug‐resistant epilepsy. We retrospectively studied a series of six SWS patients with epilepsy and extensive neurosurgical resections. At time of surgery, the patients' age ranged from 11 to 35 years (with a mean of 20.2 years). All surgical specimens were well preserved, which allowed a systematic microscopical inspection utilizing the 2011 ILAE classification for focal cortical dysplasia (FCD). Neuropathology revealed dysmorphic‐like neurons with hypertrophic cell bodies reminiscent to those described for FCD type IIa in all cases. However, gross architectural abnormalities of neocortical layering typical for FCD type IIa were missing, and we propose to classify this pattern as FCD ILAE type IIIc. In addition, our patients with earliest seizure onset also showed polymicrogyria (PMG; n = 4). The ictal onset zones were identified in all patients by subdural electrodes, and these areas always showed histopathological evidence for FCD type IIIc. Four out of five patients had favorable seizure control after surgery with a mean follow‐up period of 1.7 years. We concluded from our study that FCD type IIIc and PMG are frequently associated findings in SWS. FCD type IIIc may play a major epileptogenic role in SWS and complete resection of the associated FCD should be considered a prognostic key factor to achieve seizure control.

Introduction

In 1879, William A. Sturge 23 reported about a girl with an extensive teleangiectatic naevus of the right side of her face and head, and epileptic fits beginning in the left hand. In 1922, Frederik Parkes Weber 25 was first to radiographically demonstrate cortical calcifications in this group of patients and suggested the possibility of brain atrophy as a consequence of steal phenomena from the angiodysplastic cortex surrounding the vascular malformation. Nowadays, Sturge–Weber syndrome (SWS) is also called encephalotrigeminal angiomatosis, and classified as a neurocutaneous phacomatosis classically presenting with 1 unilateral (less frequently bilateral) facial naevus, typically located with a port‐wine patch appearance in V1 or V2 regions of trigeminal nerve innervation 2; dural and leptomeningeal angiomatosis affecting more often unilateral occipital and posterior parietal lobes 3; hemangiomas of the choroid 4; congenital glaucoma. Clinically, SWS was subdivided in three types depending on partial and incomplete manifestation into type I with facial, choroid and leptomeningeal angiomas and possible glaucoma (classic form), type II with facial angiomas but no evident endocranial involvement; and type III exclusively with leptomeningeal angiomas 21.

The association of SWS with cortical malformations including polymicrogyria (PMG) and focal cortical dysplasia (FCD) has been previously described 4, 10, 16, 18, 22. Furthermore, developmental disorganization of the cortex has been suggested as pathogenically important contribution in severe forms of SWS presenting with intractable seizures in early life 4. Whether cortical malformations exist in each SWS patient and how to classify them utilizing the ILAE classification of FCDs 2 remains to be clarified. The purpose of this study was, therefore, to systematically study histopathological changes in a series of anatomically well preserved surgical specimens obtained from six patients with SWS and intractable epilepsy, and to correlate these changes with clinical histories, neuroimaging findings and neurophysiological studies.

Material and Methods

Patients included into the study

Between 2006 and 2013, approximately 1200 surgeries for intractable epilepsy were performed at Xuanwu Hospital, Capital Medical University and 960 operations at Tsinghua University Yuquan Hospital, respectively. Six patients in both case series presented with a clinical diagnosis of SWS and were included into this study (three patients from Xuanwu Hospital, three patients from Yuquan Hospital).

Before surgery, antiepileptic drug (AED) treatment lasted for 3 months to 20 years with 2 to 3 AEDs tested, including carbamazepine, phenobarbital, diphenylhydantoin, lamotrigine, tegretol or sodium valproate, but was ineffective to control seizures. Detailed presurgical evaluation included clinical assessment, video‐electroencephalography (EEG), neuropsychological testing, brain magnetic resonance imaging (MRI), 18F‐fluorodeoxyglucose positron emission tomography (FDG‐PET) and invasive EEG studies (Figure 1). Clinical data of all patients included into this study were summarized in Table 1.

Figure 1.

Presentation of patient 1: a 35‐year‐old patient with a 34‐year history of intractable seizures. A. Axial CT demonstrated atrophy and circuitous cortical calcification with focal gyriform appearance in the left frontal lobe (arrow). B. At a level comparable with that shown in A, FDG‐PET indicated hypometabolism of the right frontal lobe (arrow). C. T1‐weighted axial MRI showing focal cerebral atrophy and a widening of the subarachnoid space in the left frontal lobe. D. The arrow points to the leptomeningeal angiomatosis. E. Intraoperative situs after surgical resection. F. Surgical specimen. Scale bar = 3 cm, applies also to G. G. Macroscopic inspection of surgical specimen after decalcification revealed extensive calcifications with focal gyriform appearance (arrows).

Table 1.

Clinical presentation of patients with Sturge–Weber syndrome

Sex	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6
	Male	Male	Male	Male	Male	Female
Age at surgery (y)	35	22	18	24	11	11
Clinical History	Negative	Febrile convulsions at 2y,callosotomy at 7y, excision left F at 10y	Negative	Negative	Negative	Negative
Disease Onset (y)	1	6	17.7	0.6	2	0.6
Intellectual deficits	No	Yes	No	Yes	Yes	Yes
Duration (y)	34	16	0.3	23.4	9	10.4
Port‐wine stain	Yes	Yes	Yes	Yes	Yes	Yes
Seizure semiology	PS	CPS	CPS	CPS	CPS	CPS
		sGS	sGS	sGS	sGS	sGS
Frequency	Weekly	Monthly	Weekly	Weekly	Monthly	Daily
Pre‐op MRI	Cerebral atrophy left F	NA	Reinforcement along pia mater in right T and O	Abnormal signal left O, low signal in T1WI and flair. High signal in T2WI.	Atrophy and dysplasia right cerebral hemisphere	Asymmetrical reinforcement along pia mater in the right cerebral hemisphere
Pre‐op CT	Atrophy calcification	Calcification	Calcification	Atrophy calcification	Atrophy calcification	Calcification
PET	HypoMet right F	NA	NA	NA	NA	HypoMet right T, P and O
Surgery	Left F	Left F, P	Right T, P, O	Left O	Right Hx	Right T, P, O
Neuropathology	SWS, IIIc	SWS, IIIc	SWS, IIIc	SWS, IIIc	SWS, IIIc	SWS, IIIc
	PMG			PMG	PMG	PMG
Seizure outcome	I (1y 2m)	I (2y 8m)	II (2y 10m)	III ( 1y 2m)	I (5 months)	na

Abbreviations: CPS = complex partial seizures; CT = comuted tomography; F = frontal lobe; Hx = hemispherectomy; HypoMet = hypometabolism; IIIc = FCD type IIIc (ILAE classification); seizure outcome according to Engel classification; NA = not available; O = occipital lobe; P = parietal lobe; PET = positron emission tomography; PMG = polymicrogyria; PS = partial seizures; sGS = secondary generalized seizures; SWS = Sturge–Weber syndrome; T = temporal lobe; y = years.

Neuropathological examination

Surgical specimens were reviewed by a neuropathologist at Xuanwu Hospital, Capital Medical University. Laboratory protocols used for specimen preparation were similar in all samples and in register to practical guidelines proposed by the Euro‐CNS Research Committee 1. Briefly, brain tissue was fixed overnight in 10% buffered formalin, orientated and cut perpendicular to the cortical surface at 5‐mm‐thick sections. Following routine paraffin embedding (Leica EG1150, Leica Microsystems, Nussloch, Germany), 4‐μm‐thin sections were stained with hematoxylin‐eosin (H&E). In addition to the review of H&E‐stained material, representative formalin‐fixed, paraffin‐embedded tissue blocs were selected for immunohistochemical procedures. The following panel of primary antibodies were used in this study (for details, see Table 2): anti‐NeuN, anti‐CD34, anti‐neurofilament, anti‐Map2, anti‐GFAP (all purchased from Beijing Zhongshan Golden Bridge Biotechnology Co. Ltd, Beijing, China). Anti‐non‐phosphorylated neurofilament (SMI32) was purchased from Covance, Emeryville, California, USA, anti‐Phospho‐S6 Ribosomal protein (pS6) from Cell Signaling Technology, Boston, Massachusetts, USA, and anti‐hyperphosphorylated tau (AT8) from Innogenetics, Ghent, Belgium. Immunohistochemical protocols followed recommendations given by the manufacturers using the streptavidin‐peroxidase (SP) technique. Briefly, an unstained slide from each case was rehydrated using xylene and graded alcohol baths. Following this, the slides were washed with tap water for approximately 5 min, washed for three cycles of 10‐min duration each. A 10‐min wash in 3% hydrogen peroxide at room temperature was carried out to prevent endogenous peroxidase activity. Autoclave sterilizer antigen retrieval followed by a 15‐min sodium citrate wash was performed for slides treated with NeuN, Map2, SMI32 and CD34 antibodies. Slides were incubated overnight in a solution of each antibody at 4°C. The slides were incubated with the manufacturer's secondary antibody and washed three times for 10‐min duration each. All slides next were incubated with the SP‐reagent, washed three times for 10 min each and incubated with 3,3′‐diaminobenzidine tetrahydrochloride for 15 min. Thereafter, slides were washed three times for 10 min each and counterstained with hematoxylin.

Table 2.

Antibodies used for immunohistochemical evaluation

Antibody (clone)	Cellular target	Dilution
NeuN (A60)	Neuronal nuclei	Prediluted
Map2 (AP18)	Neurons	Prediluted
GFAP (EP13)	Astrocytes	Prediluted
Neurofilament H Non‐phosphorylated Monoclonal antibody (SMI32)	Non‐phosphorylated neurofilaments	1:1000
Neurofilament (NF,2F11)	Neurofilament	Prediluted
Phospho‐S6 ribosomal protein antibody (pS6)	Ribosomal protein S6	1:200
Hyperphosphorylated tau(AT8)	Tau protein	1:200
CD34(QBend 10)	Vascular endothelial cell	Prediluted

Results

Clinical findings

Our study included six patients, five males and one female. At time of surgery, the patients' age ranged from 11 to 35 years (with a mean of 20.2 years). All patients had medically intractable seizures at initial examination at hospital, with age at seizure onset ranging from 7 months to 18 years (mean 4.7 years). Disease duration varied between 4 months to 34 years (mean 15.5 years). A typical facial port‐wine stain was noted in all six cases (Figure 2A). There were no family members with a neurocutaneous disorder. Four patients suffered from intellectual deficits. No patient suffered from glaucoma, hemianopsia or hemiplegia. Seizures occurred daily in one patient, weekly in four patients and monthly in one patient. Complex partial seizures and secondary generalized seizure onset were demonstrated in three, partial seizure onset in one patient. Patient 2 had a history of febrile convulsion at two years of age, dissection of corpus callosum at age 7 years, and excision of a calcified focus in left frontal lobe at age 10 years. Interictal epileptiform abnormalities occurred in all patients. Ictal EEG onset was localizing in all cases. Further details of our study population were presented in Table 1.

Figure 2.

A. Port‐wine unilateral facial naevi located in the right V1 regions of innervation of the trigeminal nerve. Axial MRI (B. T2‐weighted, C. flair) showing atrophy and dysplasia in the right cerebral hemisphere. D. HE staining revealed pseudolaminar calcification of the neocortex (arrow). E. HE staining showing cortical atrophy underneath leptomenigeal angiomatosis (arrow). F. Adjacent section to E stained with antibodies against NeuN. Note loss of cortical layers 3–5 (arrow) and microcystic degeneration (small arrow). Cortical architecture maintain altered in remote areas (asterisk), compatible with FCD type IIIc. Scale bars in D = 100 μm, E = 200 μm, applies also to F.

Imaging findings

Leptomeningeal angiomatosis and calcifications were visible on computed tomography (CT) and MRI in all cases. Cranial CT findings included intracranial dense gyriform calcifications in all cases and brain atrophy in four cases (Figure 1A). On FDG‐PET images, patients 1 and 6 showed high tracer densities within the cortex of corresponding lobes and hypometabolism in adjacent regions (Figure 1B). MR imaging findings included widening of the subarachnoid spaces in four cases (Figures 1 and 2), demonstrating cerebral atrophy. Notably, the initial hospital radiology report described cortical malformations in only two cases (patients 1 and 5).

Surgery findings

Surgical procedures were performed in all patients (left/right = 3/3). Tailored cortical resections were performed in five patients, and one patient received hemispherectomy (see Table 1). At surgery, the leptomeninges appeared thickened with numerous congested and irregularly dilated small‐caliber vessels. Pial vessels were markedly injected and showed an abnormal proliferation thereby completely covering the cerebral parenchyma with a thick, mushy, purple‐red mass (Figure 1D).

Histological findings

At macroscopic examination, brain tissue was atrophic and firm in areas covered by flaming red meningeal angiomas. Affected gyri showed marked sclerosis. Beneath the hypervascularized meninges, cortex was firm and elicited a “gritty” imprint during sectioning, owing to calcifications (Figures 1G, 2D and 3A). At microscopic level, all cases showed histopathological hallmarks of SWS with abundant tortuous and abnormal vascular structures in thickened leptomeninges, loss of cortical neurons, proliferation of astrocytes and calcifications as most significant changes (Figures 2 and 3). Calcifications were scattered throughout the cortex, often with a pseudolaminar pattern, as well as in underlying white matter. The affected cortex showed atrophy with a varying loss of neuronal layers 3 to 5 according to HE staining and NeuN, Map2 immunostaining (Figures 2E,F and 3B), and GFAP immunostaining showed astrogliosis in those atrophy areas. Layers 1 and 2 were always visible. Hypertrophic neurons with enlarged perikarya and nuclei, as well as prominent Nissl substance were frequently observed in layers 3 and 5 (Figure 4A). Some of these enlarged neurons shared cytological features of dysmorphic neurons and were immunoreactive for NF, SMI32 and pS6 (Figures 3 and 4). However, their anatomical orientation and assignment to layers 3 and 5 were still recognizable (Figure 2E,F). There were no balloon cells detectable. Immunoreactivity of phosphorylated Tau antigens was negative in all cases. CD34 immunoreactivity was only found in the vascular endothelium. In contrast to a previous report 15, the neuropathological alterations described here were classified as FCD ILAE type IIIc.

Figure 3.

Neuropathology findings in FCD ILAE type IIIc. A. HE‐stained section of a frontal ictal onset zone showing calcification in white matter and adjacent neocortex (arrow). Scale bar = 500 μm, applies also to B. B. Loss of cortical layer 3 (arrow; NeuN immunostaining) and dyslamination of deeper cortical layers compatible with FCD type IIIc. C. SMI32 immunohistochemistry revealed clusters of hypertrophic neurons in layer 3 (white arrow). Most neurons maintain, however, their anatomical orientation (black arrows). Scale bar = 100 μm. D. High power magnification of C. Scale bar = 50 μm.

Figure 4.

Hypertrophic neurons in FCD type IIIc associated with SWS. A. HE staining of FCD ILAE type IIIc revealing hypertrophic neurons with similar cytological features compared to dysmorphic neurons obtained from FCD ILAE type IIa (D–F. archival case from pathology Dept.). B. SMI32 immunohistochemistry in FCD type IIIc (B) compared with FCD type IIa (E, same patient shown in D). Note that dysmorphic neurons in FCD IIa were larger and often clustered without any anatomical orientation. F. pS6 immunohistochemistry showing marked labeling of dysmorphic neurons in FCD type IIa (same patient shown in D and E), whereas hypertrophic neurons in FCD type IIIc (C) may occur rather isolated and maintain their anatomical orientation. Scale bar in F = 50 μm, applies to all images.

Postsurgical outcome

Postsurgical seizure control was quantitatively assessed according to Engel's classification. Data were available for five patients with a mean follow‐up of 1.7 years (0.4–2.8 years). Seizure frequency improved in all the patients, with favorable outcomes (Engel class I–II) obtained in one patient after hemispherectomy and three patients after tailored resection (Table 1).

Discussion

Most children with SWS suffer from difficult‐to‐treat epilepsy. Decrease of cerebral blood flow within affected cortical areas by pial angiomatosis as well as decreased venous return, focal ischemia and decreased neuronal metabolism are considered as main pathogenetic mechanisms. However, intraoperative electrocorticographic recordings also documented epileptogenic areas beyond angiomatous patches 20. Few reports highlighted the coexistence of SWS with cortical malformations such as FCD and polymicrogyria 4, 10, 16, 18, 22. These studies reported FCD IA or FCD IIA using Palmini's classification scheme 19. In contrast, the ILAE classification of FCDs would classify architectural abnormalities with or without hypertrophic neurons outside layer 5 of the neocortex in association with any other principal lesion as FCD type III 2, that is, FCD type IIIc in SWS with extensive capillary‐venous malformations.

All of our six cases showed abnormal neurons with cytological features described also for dysmorphic neurons in FCD type II or cortical tubers. They presented with a significantly enlarged cell body and nucleus, abnormally distributed intracellular Nissl substance, and cytoplasmic accumulation of neurofilament proteins 2. Notwithstanding, FCD type II can occur as double pathology with other principal lesions by two independent pathomechanisms, that is, FCD type II and cavernomas or tumors. Herein, we propose to classify the described histopathology pattern as FCD ILAE type IIIc. A major argument is that the overall architecture of the neocortex does not fit into the concept of FCD II. Cortical thickness in FCD IIa is often hyperplastic and architectural layering severely disturbed with many dysmorphic neurons recognizable throughout the cortical thickness as well as in white matter 15. Moreover, cell shape is not a defining hallmark of dysmorphic neurons, as these cells can present with pyramidal or interneuronal phenotypes 2. Hypertropic pyramidal neurons similar to those described here and reminiscent to those observed in FCD IIa have been reported also in other epileptogenic pathologies, such as hippocampal sclerosis 6, 12, 13, 24. We propose long‐term plasticity related changes or epigenetically driven molecular pathways as alternative possibilities to convert the cytologic appearance of any pyramidal cells into a cell with dysmorphic‐like features 7, 8, rather than resulting from an unique molecular pathway distinct from SWS, that is, FCD type II. However, this hypothesis will need further exploration by clarifying the molecular nature and origin of FCD subtypes, as well as deciphering the impact of other pathologies such as ischemia and vascular calcification for aberrant morphogenesis in the developing brain 11.

Cortical malformations are frequent findings in SWS, that is, PMG, and assumed secondary to ischemia resulting from leptomeningeal angiomatosis during the second developmental trimester 4, 22, which further supports our conclusion. Other characteristic histopathological hallmarks in our case series included calcification, which could be also indicative for ischemia (ie, dystrophic calcification), but its vascular pattern (Figure 2D) more likely result from aberrant protein extravasation in SWS 11. A genetic cause of PMG associated with band‐like calcification has been reported, but was not further investigated in our series 17.

In SWS, cortical malformations are considered epileptogenic because intrinsic epileptogenicity as well as topographic relationship between the epileptogenic cortex and the FCD have been documented 14, 16. Complete resection of these FCDs is likely a prognostic key factor determining outcome of epilepsy surgery 14. If FCDs remain unrecognized in patients with focal angiomatosis, resection of the angioma and underlying cortex may result in poor outcome. Therefore, it is important to consider invasive recording modalities, anatomic and functional neuroimaging including FDG‐PET and SPECT, and validated by postsurgical histopathologic examination to localize and characterize the epileptogenic network in patients with SWS. Different strategies were proposed for surgical treatment of SWS, including cortical excisions, lobectomy or hemispherectomy owing to the unilateral and focal character of the condition in a majority of cases 5. As for cortical excisions, most resection procedures achieved a reduction of epileptic activity but followed later on by seizure relapse. It was T. Rasmussen already describing in 1972 that most active epileptogenic areas involved cortical regions adjacent to the angiomatous patches 20. Based on these observations, tailored lobectomy was considered. In our presented series, 80% of patients showed a significant benefit from tailored surgical resections. Indeed, in children with extensive hemispheric lesions and intractable epilepsy, hemispherectomy, whether anatomical or modified, is likely to be more effective. Kossoff et al reviewed all case series published in the literature reporting pediatric patients with SWS who underwent hemispherectomy procedures 9. In this meta‐analysis, 16 anatomic hemispherectomies, 14 functional hemispherectomies and two hemidecortications were performed, with hemispherectomy being most successful in this patient cohort (26/32 of the children were seizure free; 82%). It was surprising to note that an older age at surgery was positively correlated with seizure reduction. Six children with persistent seizures were operated on at a mean age of 1.3 years, whereas 26 seizure‐free children were operated at a mean age of 3.1 years. However, this result could have been partially hampered by the different number of children in both cohorts. Early onset seizures in patients with SWS typically respond less well to medical treatment and are associated with greater neurologic and cognitive impairments than late‐onset seizures 3.

Conflicts of Interest

None of the authors have conflicts 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.