Epilepsy Surgery: Outcomes and Complications

Age at Onset of Epilepsy

Patients with an earlier age (usually < 5 years) at onset of epilepsy or at time of the initial neurological insult may be up to 3 times more likely to have a favorable postoperative outcome. ^, However, some investigators proposed that this variable actually predicts HS, which is the true favorable prognostic indicator. ^, Several findings support this hypothesis: these patients were indeed more likely to have features typical of HS, such as unilateral hippocampal atrophy on MRI or focal ictal EEG with predominantly partial seizures ; and age at onset was of no prognostic value in studies evaluating pure cohorts of HS ^, or controlling for pathology. ^{, ,}

Duration of Epilepsy

A long history of seizures correlated with worse outcome in multiple studies on univariate analysis. ^{, ,} In some of those same cohorts, this influence disappeared when multivariate analysis was performed, adjusting for other, more solid indicators of outcome. ^, Furthermore, many more recent studies found no correlation of epilepsy duration with outcome. ^{, , , ,} Various hypotheses have been proposed to explain these findings, including secondary epileptogenesis occurring with a long seizure history, varying degrees of maturation of different epileptogenic foci, and increased development of generalized seizures with longer epilepsy duration.

Age at Surgery

Most studies found no correlation between age at surgery and seizure outcome, ^{, , ,} although one longitudinal study in HS patients found that patients who were 24 years or younger at surgery were about 4 times more likely to be seizure free 5 years after surgery compared with the older surgical group (≥36 years).

Absence of Secondarily Generalized Tonic-Clonic Seizures

The poor prognostic significance of secondarily generalized tonic-clonic seizures (SGTCSs) in the context of TLE surgery was confirmed in a prospective multicenter trial. Prior retrospective data leading up to this observation included a 2.2-fold increased chance of seizure freedom 5 years after surgery in patients who had no SGTCSs compared with those who did, and another study in which only 57% of mesial TLE patients with SGTCSs achieved a 1-year remission compared with an 80% remission rate in those who had only partial seizures. This effect may be most significant when generalized tonic-clonic seizures are frequent (more than two per year) and occur within 3 years of surgery.

The occurrence of SGTCSs in TLE correlates with more extensive HS, multifocal irritative areas, and extended PET hypometabolism, suggesting a diffuse potential epileptogenic zone with lower chances of postoperative seizure freedom. This hypothesis is further supported by recent data correlating early seizure spread (<10 seconds after ictal onset) outside of the anteromesial temporal lobe with a higher risk of seizure recurrence.

Low-Baseline Seizure Frequency

A lower seizure burden is correlated with more favorable seizure outcomes after TLE surgery. The cutoff point for good outcomes varies from 20 seizures per month to 30 seizures per month. A critical role for both seizure frequency and history of SGTCS in predicting outcomes after TLE surgery has been resuggested in recent work attempting to provide individualized seizure outcome prediction through a comprehensive scoring measure or a nomogram. ^,

Magnetic Resonance Imaging

The presence of a unilateral temporal lobe abnormality on MRI has been a consistently identified favorable outcome predictor. ^{, ,} Patients with MRI evidence of unilateral HS had a 54% chance of seizure freedom 10 years after ATL compared with 18% when MRIs were normal in one longitudinal study. In the prospective multicenter Epilepsy Surgery study, 75% of patients with unilateral hippocampal atrophy and a mesial temporal resection were seizure free as opposed to 55% otherwise. However, recent data suggest that such a favorable prognostic significance is actually conferred by any unilateral temporal MRI lesion, and not necessarily by HS, especially with concordant ictal and interictal EEG findings.

Favorable seizure outcomes are also possible with a normal MRI; recent data have actually shown seizure freedom rates of up to 41% to 48% as long as 8 years after ATL. ^, Although some data suggest that these patients may actually have MRI-negative or undetected HS, other studies concluded that most patients with normal-appearing hippocampi on high-resolution MRI have neocortical TLE because they have fewer febrile seizures, more delta rhythms at ictal onset, and more extensive lateral neocortical changes on PET with surgical outcomes still comparable to those of MRI-positive HS. ^, It should be emphasized, however, that surgery was successful in nonlesional patients typically when performed in the context of concordant EEG and PET data. ^{, ,} “Normal” MRIs correlating with bad outcomes in older studies using lower quality imaging may have included patients with extratemporal or contralateral pathology, findings that would currently exclude viable surgical options. ^{, ,} Lack of SGTCS and low preoperative seizure frequency correlate with more favorable surgical outcomes in nonlesional TLE, similar to lesional TLE.

Bilateral MRI lesions, including bilateral HS, reflect multiple potentially epileptogenic foci and correlate with a worse surgical outcome: 58% of patients with bilateral lesions are seizure free at 2 years compared with 78% of patients with unilateral lesions or even normal MRI. ^, Subtle hippocampal asymmetries only detected using volumetric analyses were less predictive of outcome.

Nuclear Imaging

Unilateral temporal hypometabolism on fluorodeoxyglucose–positron emission tomography (FDG-PET) is a good predictor of seizure freedom in patients with mesial TLE, independent of pathologic findings and regardless of whether MRI is normal. ^{, ,} It has been reported that 86% of patients with unilateral temporal hypometabolism ipsilateral to the side of surgery had a good outcome, as defined by more than 90% reduction in seizure frequency or Engel class I or II, with those chances slightly reduced to 82% if the MRI was normal. This number significantly dropped to 62% when PET was normal and to 50% when it showed bitemporal hypometabolism. With extratemporal hypometabolism, chances of seizure freedom are even worse: complete seizure freedom at last follow-up (mean, 6.1 years) was seen in 45% of patients with extratemporal cortical hypometabolism confined to the ipsilateral hemisphere and in only 22% with contralateral cortical hypometabolism.

Abundant data support the usefulness of ictal SPECT in localizing the epileptogenic zone in TLE, with 70% to 100% of ictal SPECT imaging being correctly localizing and only 0% to 7% incorrectly localizing. However, although the prognostic value of such localized SPECT findings is clear in extratemporal or poorly localized nonlesional temporal epilepsy, ^, its role in clear lesional TLE cases is less well defined. In a recent analysis of patients with unilateral HS visible on MRI, surgical outcome was not influenced by contralateral increased flow on ictal SPECT imaging. One hypothesis is that, owing to their low temporal resolution, ictal SPECT hyperperfusion patterns often contain both the ictal onset zone and propagation pathways. In that sense, these multilobulated hourglass-appearing patterns are best viewed and interpreted as representing the epileptic network, including both the area of ictal onset and spread, rather than aiming at identifying a single focus of maximal hyperperfusion representing the ictal onset zone. ^{, ,} Concordance between one or more subtraction ictal SPECT coregistered to MRI (SISCOM) regions of hyperperfusion with intraoperative electrocorticography (ECoG), and at least partial resection of the dominant SISCOM signal in a cohort of patients with drug-resistant focal epilepsy, provided additional useful information for predicting long-term postresection outcomes, leading the authors to conclude that such regions are likely critical nodes in more extensive, active, epileptogenic circuits.

Noninvasive Electroencephalography

Focal interictal EEG predicts a favorable outcome when lateralized to the side of surgery or when highly localized to the resected temporal lobe. Patients whose interictal EEGs showed 90% or higher predominance on the operated side had an 80% chance of complete seizure freedom after a mean 5.5 years of follow-up compared with 54% in those with lesser degrees of lateralization. In general, interictal evidence of a diffuse irritative zone predicts a worse outcome: postoperative seizure freedom is worse when interictal spiking is posterior temporal, extratemporal, or bitemporal. Bitemporal interictal spiking on surface EEG does not, however, automatically preclude postoperative seizure freedom. If 90% or more of surface interictal bitemporal spikes arise from one temporal lobe, excellent outcomes are possible (92% seizure free in the second postoperative year versus 50% if <90% lateralization). With a unilateral MRI temporal lesion, and with lateralizing noninvasive functional data, up to 64% of patients with bilateral interictal spikes achieved complete seizure freedom 1 year or later after surgery when seizure onset was strictly unilateral on invasive evaluation. Other findings consistent with unilateral HS, such as a history of febrile seizures or early onset of epilepsy (<3 to 6 years of age), also correlated with favorable outcome in patients with bitemporal interictal spikes, suggesting that contralateral spiking may simply be spread from a surgically treatable hippocampus. However, if the MRI is normal or shows widespread abnormalities, then seizure recurrence is the rule because either an extratemporal focus spreading to both temporal lobes or bitemporal epilepsy becomes more likely.

Similar concepts apply to the prognostic value of ictal EEG. Again, focal or anterior ictal EEG correlates with a more favorable outcome, and patients who had bitemporal ictal onsets on surface EEG still achieved seizure freedom rates of up to 64% at 1 postoperative year when their seizures were exclusively unilateral with depth recordings and when imaging or neuropsychological testing was also consistent with unilateral temporal dysfunction. ,

Invasive Electroencephalography

Depth electrode evaluations have traditionally been used to clarify lateralization of the epileptogenic zone in patients with suspected bitemporal or falsely lateralized TLE, whereas subdural recordings and SEEG are useful in neocortical epilepsy for extraoperative functional mapping and definition of the extent of the epileptogenic zone. Those modalities are therefore reserved for patients with a poorly defined epileptogenic zone, which may explain poorer outcomes seen in cases that required invasive recordings preoperatively compared with those that did not. ^{, ,} , Outcomes are particularly worse in patients who had prior temporal lobe resections. Yet, specific findings obtained with such invasive evaluations may provide useful prognostic information. More favorable outcomes are seen with exclusively unilateral seizure onset and ictal spiking as opposed to low-voltage fast activity, electrodecrement, or any other rhythmic sustained activity at seizure onset, whereas evolution into distinct contralateral electrographic seizures lowers seizure-free rates from 84% to 47% at 1 year. Short interhemispheric propagation times ranging from <1 second to <8 seconds, a short duration between EEG and clinical seizure onset, and diffuse or posterior temporal onset as opposed to anterior and/or middle basal temporal ictal onset are all predictors of seizure recurrence after surgery. A complex pattern of sharp transients or spikes, preceding multiband fast activity concurrent with suppression of lower frequencies on SEEG may be a “fingerprint” of the epileptogenic zone, highlighting the importance of careful analysis of the intracranial electrophysiology.