Epilepsy Surgery in Tuberous Sclerosis Complex: In Pursuit of the Epileptogenic Center(s)

Commentary

Centre of Epileptogenic Tubers Generate and Propagate Seizures in Tuberous Sclerosis.

Kannon L, Vogrin S, Bailey C, Maixner W, Harvey AS. Brain 2016:139;2653–2667.

Multiple seizure foci, seizure propagation and epileptic spasms complicate presurgical seizure localization in tuberous sclerosis. Furthermore, controversy exists about the contribution of tubers, perituberal cortex and the underlying genetic abnormality to epileptogenesis. We aimed to determine the epileptogenic substrate in tuberous sclerosis by studying spatio-temporal patterns of seizure onset and propagation on intracranial EEG recordings in which multiple depth and surface electrodes sampled multiple tubers and perituberal cortex. Ten intracranial EEG recordings (seven extraoperative, three intraoperative) from 10 children with tuberous sclerosis were analysed. Notable thickening and signal abnormality in the centre of many tubers on magnetic resonance imaging led to tuber centres being recorded with depth electrodes. Spatially-meaningful bipolar montages were reformatted incorporating channels recording only from the tuber centre, tuber rim and perituberal cortex. Interictal epileptiform discharges and ictal rhythms were analysed visually for location, field, morphology, frequency, latency and temporal dispersion. Fifteen electroclinically distinct seizures were recorded in the 10 patients. Seizure onset was recorded in tubers in all 15 electroclinically distinct seizures; in 9/10 electroclinically distinct seizures recorded with optimal spatial sampling, seizure onset was recorded in the tuber centre, with or without involvement of the tuber rim but not perituberal cortex. Quantitative electroencephalography analysis by pairwise cross-correlation confirmed that the tuber centre led the tuber rim and perituberal cortex during interictal, preictal and ictal spike trains. Seizure propagation was observed in 10/15 electroclinically distinct seizures, being tuber-to-tuber in all. Seven of the 17 tubers showing seizure propagation activated an independent ictal rhythm, morphologically distinct from that seen in seizure onset region (intra-ictal activation). Of the total 48 tubers sampled, 16 exhibited seizure onset, 17 were involved in seizure propagation and 40 exhibited interictal epileptiform discharges, 33 independent and seven propagated. Seizure onsets were recorded in 16/33 tubers with independent interictal epileptiform discharges, but 0/7 tubers with only propagated epileptiform discharges or 0/8 tubers with no epileptiform discharges (P = 0.003). Seizure onsets were recorded from 4/7 tubers with and 0/10 tubers without intra-ictal activation (P = 0.015). Thus, focal seizures and interictal epileptiform discharges in tuberous sclerosis arise in the centre of epileptogenic tubers and propagate to the tuber rim, perituberal cortex and other epileptogenic tubers. Rhythmic interictal epileptiform discharges and intra-ictal activation of propagated ictal rhythms are potential biomarkers of epileptogenic tubers. Interictal and ictal EEG features of epileptogenic tubers have similarities to focal cortical dysplasia type II, consistent with the reported imaging, histological and molecular similarities.

Epilepsy surgery can provide good seizure outcome in selected children with intractable epilepsy due to tuberous sclerosis complex (TSC) (1). There is no TSC-specific scheme for epilepsy surgery. Many factors—some patient specific and others physician/institution related—dictate schema for presurgical evaluation. Similar factors lead to variable surgical strategies: straight resection with or without intraoperative electrocorticography (I-ECoG) versus long-term extraoperative electrocorticography (LE-ECoG) using subdural grids/depths. Consequently, reported outcomes are not comparable (2–4). Most TSC epilepsy surgery series consist of patients with geographically well-dispersed discrete tubers on brain MRI, where VEEG gives good or explainable concordance, and resection of the tubers with margins provides a good seizure outcome. However, a truly difficult TSC patient for epilepsy surgery is one in whom areas delineated by presurgical evaluation are extensive or involve more than one region, somewhat concordant with brain MRI inventory of many large and small heterogeneous tuber complexes merging imperceptibly with each other, where even the intertuberal regions are subtly abnormal. In other words, the margin of one tuber from the margin of the other is not well demarcated. Epilepsy surgery is extremely challenging in such difficult patients, but it can be considered in selected cases based on the clinical situation and target seizures (most disabling or most severe partial seizures) as well as goals (when seizure freedom appears unrealistic, reduce seizure burden, reduce nocturnal seizures and risk of SUDEP, lower medication burden) for epilepsy surgery. Within this clinical context is the controversy of correlating ictal ECoG onset with the subregions within the MRI visible tubers. Whether seizures originate from the tuber center or the perituberal cortex is controversial, as is whether there is selective ictal propagation in between tubers (intertuberal spread) in addition to the surrounding cortex.

Kannan et al. claim that the tuber center is the ictal onset zone. The study findings are a descriptive analysis of ECoG in 10 TSC patients. The data were obtained in a routine clinical setting and retrospectively analyzed. Investigators correlated the grid and depth electrode contacts of interest to the subregions of MRI visible tuber, reformatted derivations to bipolar (meaningful) montages, and reanalyzed visually—and, in LE-ECoG patients--by quantitative EEG. Analyses showed consistent ictal onset in the tuber center with or without the tuber rim but never in the perituberal cortex. There have been two previous studies, and now each of the study show one of the three possibilities: In first study by Major et al., the findings were opposite. ECoG of three TSC patients showed the electrodes in the center of the tuber were silent, and the ictal onset zone was from the surrounding (perituberal) cortical tissue (5). In second study by Ma and colleagues, results were heterogeneous (6). ECoG of 12 TSC patients and 309 recorded seizures demonstrated heterogeneity in the ictal onset as well as the involvement of the tuber and perituberal cortex within the same and between different patients without a definite subtuberal region ictal onset. While MRI definition of the tuber center and tuber rim is relatively simpler to grasp, there is no accepted definition of perituberal cortex. In the two previous studies (5–6), the authors loosely defined perituberal cortex as the “adjacent” cortex surrounding the tuber. In the Kannan study, perituberal tissue was defined as “normal appearing surrounding cortex in the adjacent gyri around the tuber.” Although it is unclear, the tuber rim of the current study could be the perituberal cortex of the previous two studies (5–6), which may account for divergent conclusions. Distinction between the tuber rim and the perituberal cortex may not be feasible in truly difficult TSC patients, as the tubers merge with the surrounding cortex and other tuber complexes imperceptibly, and “tuber rim” and “normal perituberal tissue” cannot be precisely drawn in three dimensions on a routine brain MRI.

This retrospective study has other issues as well. There is a hierarchical bias in selecting 10 TSC patients out of 66 who underwent 101 epilepsy surgeries. Inclusion criteria were non-specific and were developed and applied in retrospect. ECoG was acquired under varying conditions: Eight patients had grid and depth electrodes, and two had only subdural grids. Seven had LE-ECoG and three had short term I-ECoG done under anesthesia. Six patients had failed epilepsy surgery, and it is possible that their ECoG interpretations could have been impacted by decorticated white matter from previous resections (chronic periodic spikes) and aberrant ictal propagation in the vicinity of previously resected tubers. Depth electrode placement was visual under neuronavigational guidance rather than robotic and stereotactic, and multiple passes in the tuber center were often performed, correlated later with brain MRI. Seizure outcome was seizure freedom in four and >90% seizure reduction of the “target seizure” in one. In five with <75% seizure reduction, many caveats were noted. In summary, their results are far from convincing and fail to settle the controversy regarding ictal potential of the tuber center versus tuber periphery.

The study reaffirms findings that are previously well recognized. First, periodic/semiperiodic sharp waves are a useful marker of epileptogenic tubers in patients who have not undergone previous resections. Second, seizure semiology in young and cognitively impaired children is often bland (staring, behavior arrest) during partial seizures, and clinical onset and offset may be difficult to recognize. Ictal motor semiology usually reflects ictal spread rather than ictal onset and, therefore, these patients should not be rejected as surgical candidates based on cursory analyses. Further, two dissimilar ictal EEG patterns can appear simultaneously or propagate in two geographically distinct tuber regions, suggesting intertuberal connectivity to generate or propagate seizures. This is commonly seen across the frontal and parietal tuber complexes (frontoparietal network). Distinguishing the primary ictal origin can be difficult even on LE-ECoG, and often both frontal and parietal tubers are resected.

In summary, epilepsy surgery in TSC remains challenging and should be driven by a sound presurgical hypothesis based on the objective analysis of clinical circumstances and surgery goals, semiology, scalp VEEG analysis, and imaging tools. Prudent use of LE-ECoG and I-ECoG could be considered especially in truly difficult patients to confirm or refute the preimplantation epileptogenic zone/network hypotheses and, in some instances, to salvage eloquent function.

For clinical care, distinguishing epileptogenicity in the tuber center versus periphery has little relevance in sparse tuber cases, as in the end the entire tuber with center and margins is resected. Clinical relevance may exist in difficult patients, as identification of epileptogenic “tuber subregions” may better define surgical strategy or even help develop noninvasive pre-surgical markers on high resolution MRI-based techniques or new PET ligands with the ultimate goal of improving epilepsy surgery outcome. Only well-orchestrated and prospectively designed studies with clinical, imaging, and basic science techniques may answer this question.