Article Text

Download PDFPDF

Epilepsy surgery
  1. Fergus Rugg-Gunn,
  2. Anna Miserocchi,
  3. Andrew McEvoy
  1. Department of Clinical and Experimental Epilepsy, University College London Institute of Neurology, London, UK
  1. Correspondence to Dr Fergus Rugg-Gunn, Dept of Clinical and Experimental Epilepsy, University College London Institute of Neurology, London WC1N 3BG, UK; f.rugg-gunn{at}


Epilepsy surgery offers the chance of seizure remission for the 30%–40% of patients with focal epilepsy whose seizures continue despite anti-epileptic medications. Epilepsy surgery encompasses curative resective procedures, palliative techniques such as corpus callosotomy and implantation of stimulation devices. Pre-surgical evaluation aims to identify the epileptogenic zone and to prevent post-operative neurological and cognitive deficits. This entails optimal imaging, prolonged video-electroencephalogram (EEG) recordings, and neuropsychological and psychiatric assessments; some patients may then require nuclear medicine imaging and intracranial EEG recording. The best outcomes are in those with an electro-clinically concordant structural lesion on MRI (60%–70% seizure freedom). Lower rates of seizure freedom are expected in people with extra-temporal lobe foci, focal-to-bilateral tonic-clonic seizures, normal structural imaging, psychiatric co-morbidity and learning disability. Nevertheless, surgery for epilepsy is under-used and should be considered for all patients with refractory focal epilepsy in whom two or three anti-epileptic medications have been ineffective.

  • epilepsy
  • epilepsy surgery

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Approximately 60%–70% of people with focal epilepsy become seizure free with medication. However, in those 30%–40% whose seizures continue despite medication, clinicians should consider other options, such as epilepsy surgery, vagus nerve stimulation or ketogenic diet. Surgical success may be defined as the complete cessation of seizures without post-operative cognitive, psychiatric or neurological dysfunction. The outcome of surgery typically represents a balance between seizure control and post-operative deficit. The pre-surgical investigative pathway aims to inform this process by localising the epileptogenic zone, eloquent cortex, and major white matter tracts.

Refractory epilepsy is the failure to achieve sustained seizure freedom despite adequate trials of two tolerated, appropriately chosen and used anti-epileptic drug schedules, either as monotherapy or in combination.1 Surgical treatment of refractory temporal lobe epilepsy improves both the seizure outcome (58% seizure free) and the quality of life, compared with optimal medical management (8% seizure free).2 Attaining seizure freedom is also associated with reduced mortality, for example, from sudden unexpected death in epilepsy.3

About 1.5% of people newly diagnosed with epilepsy may eventually require epilepsy surgery. In the UK, this equates to approximately 450 patients per year; the number of operations performed equals the number of emergent new cases. However, there are also many people with drug-refractory epilepsy who might benefit from surgery but are not considered or assessed.

The delay between onset of focal epilepsy and epilepsy surgery is still 15–20 years,4 5 indicating delay and under-referral for this potentially life-changing treatment.

Most patients with refractory focal epilepsy can be appropriately considered for surgery. Over half of those referred will not be suitable for current resective options, but they may be suitable for a palliative procedure or implantation of a stimulator.

In general, about 50% of those who undergo initial, non-invasive, pre-surgical investigations do not proceed further; 25%–40% are offered a resection without needing further investigations, and 10%–30% require intracranial electroencephalogram (EEG) recordings.

Who should be considered/referred for epilepsy surgery?

  • All adults with refractory focal epilepsy (adequate trials of two or more appropriately selected anti-epileptic medications) should be considered for resective/curative surgery

  • All adults with refractory focal or generalised epilepsy who are not candidates for resective surgery should be considered for palliative procedures, such as vagus nerve stimulation

  • A low IQ or memory impairment is not a contraindication to resective surgery

  • Older patients should be considered for surgery but the risk of complications is higher

  • A history of long-term psychiatric disorder does not exclude a patient from resective surgery but close psychiatric supervision post-operatively would be mandatory.

  • Bilateral interictal epileptiform activity is not a contraindication to resective surgery; unilateral onset seizures often have bilateral interictal epileptiform activity

  • People with refractory focal epilepsy and a normal structural MRI scan of brain should be considered for surgery; other investigations may identify a single epileptogenic zone amenable to surgical resection

  • Multiple or diffuse lesions on MRI are not a contraindication to surgery; seizures may arise from only one of the visible abnormalities or from a part of the lesion.

  • People whose seizure semiology suggests involvement of primary eloquent cortex should be considered for surgery; essential functions can be localised and the symptomatic zone may be distinct from the epileptogenic zone due to seizure propagation.

Note that all patients with refractory focal epilepsy and an apparently clinically concordant lesion on initial imaging still need comprehensive pre-surgical evaluation; a lesion’s epileptogenicity must always be confirmed before resection. The exception is when the primary rationale for surgery is tumour resection, rather than seizure control.

Pre-surgical evaluation

The principal aim of pre-surgical evaluation is to determine the epileptogenic zone and its relationship to eloquent areas of the brain. The epileptogenic zone is a theoretical construct, defined as the minimum amount of cortex that must be resected (inactivated or completely disconnected) to give seizure freedom. No single pre-operative investigation can characterise the epileptogenic zone completely reliably, and even when combining various investigative modalities there may still be variable concordance.(figure 1)

Figure 1

Sir Victor Horsley (far left) in the operating theatre at Queen Square, 1906. Accompanied by Dr Powell, anaesthetist and Professor Theodor Kocher from Berne (far right), a strong advocate of asepsis during surgery.

When pre-operative non-invasive investigations have a high degree of congruence between these zones, it may be possible to recommend surgery with predictable levels of benefit and risk. However, if non-invasive investigations are discordant, proceeding directly to resective surgery may be rejected in favour of establishing more definitive localising data using, for example, invasive EEG recordings (figure 2).

Figure 2

Epilepsy surgery assessment (reproduced from Duncan JS and colleagues by permission of The Lancet).34 EEG, electroencephalogram; ¹⁸F-FDG, ¹⁸F-fluorodeoxyglucose; fMRI, functional MRI; MEG, magnetoencephalography; PET, positron emission tomography; SPECT, single-photon emission CT.

Identifying the seizure focus: non-invasive techniques

Identifying the epileptogenic zone starts with a detailed elucidation of the nature and tempo of the seizure semiological characteristics. Initial investigations include detailed imaging and interictal and ictal EEG recordings (figure 2). High-resolution anatomical MRI with an epilepsy-dedicated protocol should be performed (ideally on a 3-Tesla MRI scanner) and evaluated by an expert neuroradiologist. Up to 86% of cases of hippocampal sclerosis may remain undetected using a standard MRI sequence reported by a non-expert radiologist.6 A 3-Tesla MR scanner improves the identification of structural lesions by up to 20% compared with a 1.5-Tesla scanner (figure 3).7

Figure 3

1.5T (left) and 3T (right) axial T2-weighted MR images showing left frontal focal cortical dysplasia. The lesion is more readily identifiable on the 3T-derived images.

Prolonged video-EEG telemetry is mandatory, often with anti-epileptic drug reduction to increase the number of seizures recorded within a reasonable time frame. During and immediately after seizures, it is important to perform neurocognitive testing to establish functional deficit, to aid localisation.8 It is not appropriate to construct a surgical hypothesis using only interictal data as this only poorly defines the epileptogenic zone.

If the initial investigations are discordant or, if the MRI scan of brain is normal or non-definitive, nuclear medicine studies such as fluorodeoxyglucose-positron emission tomography (figure 4) and ictal single-photon emission CT (figure 5) are used to generate a hypothesis that may then be tested with intracranial EEG recordings. Additionally, more advanced developmental MRI or neurophysiological techniques—such as magnetoencephalography (figure 6) or electrical source imaging—may help to localise the seizure focus.

Figure 4

¹⁸FDG PET scan showing left temporal hypometabolism in a 32-year-old man with left temporal lobe epilepsy and a normal MRI scan of brain (A, axial slice; B, coronal slice). (C) Surface rendered left lateral projection of glucose uptake. (D) Results of a statistical voxel-based comparison of surface-rendered glucose uptake in the patient, compared with a set of control data (using Neurostat-3D SSP software). There is an area of hypometabolism (green) in the left temporal lobe (D). The patient became seizure free after left temporal lobe resection (adapted from Rathore and colleagues by permission of Elsevier35). FDG-PET, fluorodeoxyglucose-positron emission tomography.

Figure 5

Ictal SPECT scan in patient with temporal lobe epilepsy and previous middle cerebral artery territory infarct with hemiparesis. Coronal T1-weighted MR image showing areas of post-infarction encephalomalacia (left) and with superimposed subtraction SPECT perfusion image (middle). This shows an area of ictal hyperperfusion in the left temporal lobe, when the ictal perfusion scan is subtracted from the interictal SPECT scan (right). SPECT, single-photon emission CT.

Figure 6

MEG informing the placement of intracranial EEG electrode placement. (A) axial T2-weighted MR image of 30-year-old man with pharmaco-resistant frontal lobe epilepsy and normal structural imaging. MEG dipoles reconstructed from interictal epileptiform activity detected with a 275-channel whole-head magnetometer (E) are superimposed on axial (B), sagittal (C) and coronal (D) T1-weighted MR images and localise to the anterior cingulate. This was evaluated with a mesial frontal strip of intracranial EEG electrodes (F) and seizure onset (red circles) co-localised to the MEG dipoles. The patient subsequently underwent resection (post-operative axial T2-weighted MR image (G) and remains seizure free for 5 years. Histopathological evaluation identified focal cortical dysplasia. EEG, electroencephalogram; MEG, magnetoencephalography.

Identification of the seizure focus: invasive techniques

Up to 20%–30% of surgical candidates need intracranial EEG recordings to define the epileptogenic zone. The aim of invasive EEG recording is to acquire neurophysiological data to support or disprove a hypothesis regarding the site of seizure onset. Typically, this is required for non-lesional focal epilepsy or if non-invasive investigations are non-localising or discordant (figure 2).

The type of intracranial recording depends on the suspected pathophysiological substrate of the epilepsy and its location. Intracranial EEG recording mainly involves two techniques:

  • Depth electrode implantation uses multiple electrodes stereotactically implanted into the brain parenchyma via small screws fixed to the skull (figures 7–9). This stereo-EEG technique allows recording from both deep and superficial areas. Its overall morbidity is about 1.3%, equating to a risk of 1 in 287 electrodes, including a 1% risk of haemorrhage.9 The implantation can be performed with a frame-based stereotactic approach or using frameless image-guidance systems that promise to simplify the pre-surgical planning of electrode placement (figures 8 and 9). Both frame-based and frameless approaches can be robotically assisted.

  • Subdural electrodes (strips and grids) are placed directly on to the brain surface (figures 10–12). Subdural strips can be placed through simple burr holes, whereas grids require a craniotomy and can record from a larger area of contiguous cortex. They are frequently used when epileptogenic lesions are adjacent to eloquent cortex. Subdural grids enable detailed extra-operative direct cortical stimulation, facilitating the mapping of eloquent cortex. The main advantage of this technique is a more comprehensive cortical stimulation study, compared with stereo-EEG, where cortical sampling is spatially more limited.

Figure 7

Three-dimensional surface-reconstructed MR image (left) and perioperative photograph (right) showing placement of depth electrodes as part of a stereo electroencephalogram study.

Figure 12

Post-operative MRI scan of brain coronal T1-weighted images (top row) and corresponding skull radiograph (bottom row) showing placement of subdural intracranial electroencephalogram recording electrodes.

Figure 11

Intra-operative photographs and reconstructed three-dimensional images: before placement of subdural electroencephalogram grid (left) and after placement (right).

Figure 10

Intra-operative photographs showing placement of subdural electroencephalogram recording strips and grids.

Figure 9

Axial (top left), sagittal (top right) and coronal (bottom left) T1-weighted MR images and three-dimensional surface reconstructed image (bottom right) with superimposed depth electrode electroencephalogram proposal planned with neuronavigation software, before surgery. Electrode entry point, trajectory, depth and target localising information can be uploaded to the surgical navigation software during surgery to guide accurate and safe placement (courtesy of Vejay N Vakharia, UCL Institute of Neurology).

Figure 8

Three-dimensional surface-reconstructed MR image with superimposed vasculature derived from digital subtraction or CT angiography and pre-operative depth electrode EEG placement strategy for stereo EEG study using neuronavigation software. Each EEG electrode is depicted in a different colour (courtesy of Vejay N Vakharia, UCL Institute of Neurology). EEG, electroencephalogram.

The duration of invasive monitoring depends on the seizure frequency, the success of any planned stimulation and patient compliance. It may need several weeks of depth electrode recording to characterise a patient’s seizure disorder fully. In contrast, subdural grid recordings seldom extend beyond 10–14 days, as these patients often have a higher seizure frequency, and the procedure carries a higher risk of infection.

Invasive monitoring may be stopped at any stage if there is a clinically significant adverse event, such as intracranial haematoma (<5% of cases) or infection resulting from the wires passing through the scalp (2% of studies).10 These risks can be reduced by careful intra-operative technique, appropriate post-operative nursing care and prophylactic antibiotics. Invasive intracranial EEG studies are time-consuming, expensive, have an inherent risk of complications, and require numerous personnel and access numerous allied investigations. This limits the number of neuroscience centres that can support a comprehensive epilepsy surgery programme.

About 40% of patients who undergo invasive recording are deemed unsuitable for resective surgery, for three main reasons: the epileptogenic zone cannot be satisfactorily determined, there are multiple potential seizure foci or the epileptogenic zone is situated in eloquent cortex.

Preservation of cognitive and neurological function

Neuropsychological testing can help to predict the post-operative cognitive outcome and seizure control.11 Whereas bilateral hippocampal resection results in profound anterograde amnesia, unilateral temporal lobe resections may result in material-specific memory dysfunction. Approximately 30% of patients undergoing dominant temporal lobe resection develop difficulties in verbal memory processing and word-retrieval; a similar proportion of patients undergoing non-dominant temporal lobe resection develop difficulties with non-verbal or visual memory processing. A decline in verbal memory is generally more disabling than a decline in visual memory. Those at greatest risk of a troublesome decline in language and verbal memory are high-functioning people who undergo an anterior temporal lobe resection in the speech-dominant hemisphere. A proportion of people note an improved memory after temporal lobe resection, particularly if the resection is on the non-dominant side.

Pre-operative neuropsychological scores, in conjunction with MRI and other clinical data, can be used to predict post-operative neuropsychological change using logistic regression techniques.12 Those at high risk of a significant memory decline can be advised pre-operatively and can be trained in compensatory strategies before the surgery.

Functional MRI (fMRI) can lateralise and localise cerebral areas involved in language function. Language lateralisation assessed using fMRI language tasks correlates well with that assessed using the carotid amytal test.13 Thus in most epilepsy surgery centres, language fMRI has largely replaced the amytal test.

Resections close to eloquent language cortex require a more detailed and accurate assessment of the anatomical relationship between seizure focus and language areas than can be assumed from an fMRI activation pattern; it may be necessary to perform electro-cortical stimulation or an awake resection (figure 13). Awake craniotomy poses significant challenges to both surgeon and anaesthetic team; seizures may occur during the procedure or patients may become agitated or distressed. There is neither immediate control of blood pressure nor the possibility of hyperventilation to reduce intracranial pressure.

Figure 13

Intra-operative photographs showing the use of cortical stimulation (left) to identify eloquent cortex and (right) post-operative surgical field. Note the preservation of branching vasculature.

Tractography derives from diffusion-tensor imaging; it allows identification of nerve fibre tracts within the brain, and demonstrates the structural basis of connectivity between brain regions. Visual pathway tractography can predict the likelihood and extent of a visual field defect following anterior temporal lobe resection. Typically, up to 10% of patients undergoing anterior temporal lobe resection develop a significant superior quadrantanopic field defect post-operatively, which can prevent restoration of driving eligibility (figure 14). Tractography data made available to the surgeon through a visual overlay when using the operating microscope has reduced the incidence of post-operative visual field deficit.14

Figure 14

Post-operative coronal (top left) and sagittal (top right) T1-weighted MR images with superimposed pre-operative Meyer’s loop optic radiation derived from diffusion tensor imaging tractography. There is anterior extension of Meyer’s loop into the temporal pole, which resulted in a quadrantanopic visual field defect, a recognised complication of anterior temporal lobe resection. Intra-operative photographs (bottom row) through the surgical microscope with optic radiation tractographic information (bottom right) superimposed to inform the surgeon and reduce the risk of post-operative field defect.

Epilepsy surgery multidisciplinary meeting

A multidisciplinary and systematic approach to investigations is essential to a surgical pathway (figure 2). During the epilepsy surgery multidisciplinary team meeting—attended by epileptologists, neurosurgical team, neurophysiologists, neuropsychologists, neuropsychiatrists and epilepsy specialist nurses—the seizure history is presented and then each investigation discussed in detail. Having reached consensus on a potentially curative or palliative surgical approach, the team formulates a detailed management plan. Subsequently, this risk:benefit analysis is discussed in detail with the patient and along with written information (Box 1). Patients must be given realistic expectations of what may be achieved, and what may be the negative consequences of epilepsy surgery(Box 2).

Box 1.

Examples of pre-surgical discussions with patients

Patients with refractory focal epilepsy and electroclinically concordant hippocampal sclerosis

  • ‘We have now had a chance to review the results of all of your investigations in the surgery multidisciplinary meeting. Your MRI brain scan has shown evidence of scarring in the inner aspect of the left temporal lobe, called hippocampal sclerosis, and this is a common finding in people with difficult-to-control seizures. The brain-wave EEG recordings have confirmed that your seizures start from this area and the memory and language tests performed by the neuropsychologists showed that your memory for words is less efficient than your memory for pictures and places. Again, this suggests that the left temporal lobe is the region of your brain from which your seizures arise.

  • It was the consensus of the meeting that we can offer surgery to help your epilepsy. Specifically, we propose removing part of the temporal lobe including the area of scarring. We estimate that by doing so, there is a 60%–70% chance of stopping your seizures with an additional 15% chance of significantly improving your seizures but without stopping them completely. There is a 1 in 100 chance of your seizures worsening after surgery.

  • Of course, no operation on the brain can take place without risk. I would estimate that there is a 1 in 100 chance of developing a serious new problem, such as weakness of an arm or leg or difficulty with speech and this may not be recoverable. The risk of a less severe but noticeable problem such as loss of vision in the top right hand corner of both eyes that would prevent you from driving even if you became seizure free is about 1 in 10. There is a 1 in 3 chance that your memory for words may not be as good following surgery but if the seizures are stopped we would expect this to level out over time. Also, about 1 in 3 people undergoing surgery of this kind experience mood disturbance afterwards but this is typically temporary and improves after 3–6 months. Whilst the risks may seem worrying, it is important to consider that the risks of surgery are very similar to the risks you run from your epilepsy, as it currently is, over a 2-year period. Furthermore, medications alone are associated with a less than 5% chance of stopping your seizures, and if seizure continue, it is likely that your memory will worsen over the coming years’.

Patients with refractory focal epilepsy and normal optimal imaging requiring an intracranial EEG study.

  • ‘We have now had a chance to review the results of all of your investigations in the surgery multidisciplinary meeting. Your MRI brain scan appears entirely normal but the PET scan, which looks at the amount of sugar the brain uses, suggests that your seizures may be arising from the front of the brain on the right hand side. The brain-wave EEG recordings suggest that your seizures start from this area and the memory and language tests performed by the neuropsychologists are consistent with this.

  • It was the consensus of the meeting that we are not able to proceed directly with surgery to help your epilepsy but that further investigations are required to see whether this will be possible in due course. Specifically, we propose performing a surgical operation to place EEG recording electrodes inside the brain, through small holes in the skull. This is called intracranial EEG. The aim of this is to find out very accurately which area of the front of the brain gives rise to the seizures and to see if an operation to remove this area would stop your seizures. We estimate that there is a 50%–60% chance of the intracranial EEG test identifying an area of the brain from where your seizures arise. If this is removed at an operation afterwards, we estimate that there is a 50% chance that your seizures would stop. This means that if we proceed at this point, you have a 30% chance overall of becoming seizure free with surgery.

  • Of course, no operation on the brain can take place without risk. I would estimate that as a result of the operation to place electrodes inside the brain, there is a 2–3 in 100 chance of developing a serious new problem, such as a bleed inside the brain. This may lead to symptoms similar to a stroke including weakness of an arm or leg and a second emergency operation might be needed. There is also a 2–3 in 100 risk of infection. If one of these major complications of surgery arose the EEG recordings might need to be stopped early, even if useful information about your seizures had not yet been obtained.

  • Should you go on to have surgery to remove a part of the brain that we think is giving rise to your seizures, I would estimate that there is a 1 in 100 chance of developing a serious new problem, such as weakness of an arm or leg or difficulty with speech and this may not be recoverable. There may be other risks of developing a new problem depending on exactly what operation is performed and this will be discussed with you in detail before proceeding.

  • Whilst the risks may seem worrying, it is important to consider that the risks of surgery are very similar to the risks you run from your epilepsy, as it currently is, over a 2-year period. Furthermore, medications alone are associated with only a 5% chance of stopping your seizures’.

Box 2.

Benefits/risks of surgery

Potential benefits

  • Seizure freedom2 25

  • Reduced seizure severity36

  • Reduced medication load37

  • Cognitive gains from reduction of both medication load and seizure activity38

  • Reduced risk of sudden unexpected death in epilepsy (SUDEP) and injury3 39

  • Possible improved long-term psychiatric outcomes40

  • Improved quality of life41


  • Perioperative mortality and morbidity26

  • Post-operative neurological and cognitive deficits42 43 particularly if seizures continue post-operatively42

  • Possible short-term and de novo long-term psychiatric complications44 45

Surgical Procedures: curative


The increased anatomical resolution of modern MRI has led to the identification of many more cortically based lesions. Small lesions such as cavernomas, focal areas of cortical dysplasia (figure 15) and indolent tumours such as dysembryoplastic neuroepithelial tumours are highly epileptogenic, and their resection is associated with a high rate of seizure freedom.

Figure 15

Coronal fluid-attenuated inversion recovery MR images showing right frontal lobe focal cortical dysplasia with typical T2 hyperintense tail towards the ventricle seen in focal cortical dysplasia type IIb.

Interventional MRI allows documentation of complete lesion resection before completing the surgical procedure. It also enables the surgical navigation software to be recalibrated during the operation, improving its accuracy. Thus, interventional MRI can potentially improve the rate of seizure freedom from surgery, and reduce the risk of neurological deficit (figures 16 and 17).

Figure 16

Interventional MRI operating suite. The red line represents the 5-gauss static magnetic field strength around the scanner.

Figure 17

Interventional MRI. Intra-operative MR images showing desired area of resection (purple outline) and corticospinal tracts derived from diffusion tensor imaging tractography (orange outline) superimposed on intra-operative MR images. The pre-operative (left cluster) and perioperative (right cluster) investigations show the extent of resection and post-operative preservation of corticospinal tracts.


Temporal lobe

Anterior temporal lobe resection, including of the mesial temporal lobe structures, accounts for about half of operative procedures performed in specialist epilepsy centres. Contemporary approaches attempt to limit the size of the neocortical resection to minimise neurocognitive sequelae, using either the method described by Spencer15 or selective amygdalohippocampectomy. Anterior temporal lobe resection that includes removal of up to 4.5 cm of neocortex shows a trend towards improved seizure outcomes compared with selective amygdalohippocampectomy, and with minimal differences in neuropsychological outcome. Despite this, there is still controversy about the different approaches. Having familiarity with a specific approach or technique is associated with improved seizure outcome and a lower morbidity, and this should therefore influence the surgical strategy.

The initial seizure-free rate following resection of hippocampal sclerosis (figure 18) is approximately 75%–80%, and approximately 70%–75% for resection of other temporal lobe lesions.4 There is gradual attrition of seizure freedom over subsequent years so that 40%–50% can expect to have remained totally seizure free after 20 years. The failure to achieve seizure freedom may be due to insufficient resection of the epileptogenic mesial temporal structures, seizures arising from the contralateral mesial temporal lobe, lateral temporal neocortical epilepsy, dual pathology, secondary epileptogenesis and extra-temporal lobe epilepsy mimicking temporal lobe epilepsy, as with insula or posterior cingulate foci.

Figure 18

Coronal T1-weighted (left) and fluid-attenuated inversion recovery (right) MR images showing left mesial temporal lobe sclerosis.

Extra-temporal lobe:

Seizure outcome is typically less good for extra-temporal resections than with temporal lobe resections, particularly in non-lesional cases4 but may be as high as 50%–60% following careful evaluation with stereo EEG.16 Depending on the pathology, it may be necessary to make large resections of the epileptogenic zone, thus putting eloquent cortex at risk. Nevertheless, pre-operative discussion may have determined a patient preference for risking neurological deficit rather than having persistent seizures. Certain post-operative neurological and neuropsychological deficits are often predictable, such as a hemianopic field deficit following occipital lobe resections, or a Gerstmann’s syndrome or hemisensory impairment following parietal lobe surgery. The extent and location of the resection depends upon the presence and extent of a structural lesion, the location of eloquent cortex as defined by fMRI or electrical stimulation studies, the position and trajectory of white matter tracts, and the location of major blood vessels.


Hemispherotomy inevitably leads to profound neurological deficit, including hemiplegia and hemianopia; this is therefore most appropriate for those with pre-existing deficit (figure 19). If patients can walk pre-operatively, most remain able to do so. Typically, there is loss of fine motor skills in the contralateral upper and lower limbs, but cognitive function is stable. Hemispherectomy used to cause long-term complications in up to one-third of patients, including superficial cerebral haemosiderosis. As a result, alternative techniques were developed, including a functional hemispherectomy, in which the temporal lobe and central cortex are removed and the corpus callosum and frontal and occipital cortex disconnected. The seizure outcomes remain unchanged, but the complication rate has significantly improved.

Figure 19

Axial fluid-attenuated inversion recovery (left) showing left-sided Rasmussen’s encephalitis and coronal T1 (right) MR images showing left hemispherectomy.

The success of hemispherectomy depends on the underlying pathology, with excellent outcomes and seizure freedom rates approaching 75%–85% for pathologies such as Rasmussen’s encephalitis and focal infarcts, but with a poorer outcome for patients with hemi-megalencephaly.17

Surgical procedures: Palliative

The objective of these functional procedures is to palliate rather than to cure the epilepsy. They should be offered only if resective surgery is considered to be inappropriate or too risky.

Corpus callosotomy

The primary indication for corpus callosotomy is atonic drop attacks, although it is effective for other seizure types. The disconnection slows inter-hemispheric seizure propagation and provides patients with a warning or disrupts a seizure, the expression of which relies on synchrony . About 74% of people have favourable outcomes with corpus callosotomy, including 39% who stop having drop attacks.18 The operation may cause either immediate or delayed symptoms of disconnection. In order to minimise this risk in adults, the callosotomy is usually carried out in two stages, with the anterior two-thirds of the corpus callosum being divided first and the posterior third divided later (figure 20).

Figure 20

Coronal T1-weighted (left) and axial diffusion tensor imaging tractography (right) MR images showing anterior corpus callosotomy with interruption of anterior but not posterior callosal tracts.

Stimulation techniques

Vagus nerve stimulation

The pathophysiological basis of periodic vagus nerve stimulation has not been fully elucidated but may involve autonomic nervous pathways and augmented function of neurotransmitters such as gamma-aminobutyric acid (GABA). Besides intermittent stimulation, the patient or companion can also effect on-demand stimulation. The newest devices also stimulate on detection of ictal tachycardia.

The left vagus nerve is used to avoid cardiac side effects, and the electrode placed on the nerve in the neck between the common carotid artery and the internal jugular vein. Side effects include hoarseness and coughing during stimulation and neck discomfort. The beneficial effect of vagus nerve stimulation may take up to 2 years to emerge. Long-term studies have shown that although very few people become seizure free, up to 43%–64% have their seizure frequency reduced by 50% or more.19

Intracranial stimulation

Traditional ‘open-loop’ deep-brain stimulation techniques use continuous or scheduled stimulation, and so do not depend on the presence of epileptiform activity. In one study of people with focal epilepsy, bilateral stimulation of the anterior nuclei of the thalamus was associated with an immediate mean decrease in seizure frequency of 29%, and of 56% at 2 years. The procedure was generally well tolerated without symptomatic haemorrhage or infection, but the treatment group developed more depression and memory difficulties.20

‘Closed-loop’ or responsive cortical stimulation has been an important development. Here, the stimulation is provided only having detected abnormal ictal or interictal epileptiform discharges. The Neuropace responsive neurostimulator (RNS) delivers a pre-symptomatic short train of electrical pulses to the brain through implanted leads, on detection of abnormal electrical activity via an implanted strip electrode on the brain surface. A multicentre, double-blind, randomised control trial showed a 38% reduction in mean seizure frequency in the treatment group compared with a 17% reduction in the sham group.21 As with vagus nerve stimulation, the efficacy improved over time, with a 66% seizure reduction at 6 years. Mild adverse events, such as implant site pain, headache and dysaesthesia, were common in both the treated and sham groups.

Other implantable responsive devices currently in development use optogenetics, local cooling or drug-delivery systems; trials of these in humans will begin in the next few years.


Seizure control

The outcome from epilepsy surgery is based on several facets including seizure control, neuropsychological development, neurological deficit, quality of life and psychosocial adjustment. Of these, seizure control is the one most commonly ascertained. In one large cohort study of almost exclusively curative procedures, the average post-operative seizure remission rate was 52% at 5 years and 47% at 10 years, with a range of between 40% for extra-temporal lesionectomy and 64% for hemispherectomy.4

Early seizure recurrence predicted a worse seizure outcome, in line with other studies.22

Patients with an identified epileptogenic lesion are two to three times more likely to become seizure free post-operatively than patients with normal imaging.23

The factors associated with an increased risk of seizure recurrence post-operatively include normal MRI scan of brain, a history of focal-to-bilateral tonic-clonic seizures, a psychiatric history, extra-temporal rather than temporal lobe surgery, older age, and having tried a higher number of medications before surgery (figure 21).24

Figure 21

Probability of seizure freedom in selected groups. The group with the best chance of seizure freedom (HS), no SGTCS, temporal surgery, no psychiatric history and no learning disability) compared with single significant prognostic features. (Reprinted with permission.24 (2017). BMJ. All rights reserved.) HS, hippocampal sclerosis; SGTCS, secondarily generalised tonic-clonic seizures.


The neurological complications of epilepsy surgery depend largely on the location and extent of the surgical resection. The overall complication rate is around 7%–8%,25 though this is higher in people aged over 50 years at 6%–25%. Most neurological deficits are predictable (such as visual field deficit); the risk of new, long-term, unexpected neurological complications is low, at less than 5%. Transient complications such as infection or cerebrospinal fluid collections are more common (figure 22).26

Figure 22

Coronal T1 MR image (left) showing right inferior temporal lobe encephalocoele (circled) with intra-operative photograph of defect in floor of middle cranial fossa (middle). Post-operative coronal T1-weighted MR image showing evidence of right temporal lobe resection and two complications of surgery—a right-sided subgaleal cerebrospinal fluid collection and a left-sided septated subdural haemorrhage.

Psychiatric disorders are common in people with epilepsy. Having a prior or current history of psychiatric disorders is associated with a lower chance of seizure freedom following surgery, but this is not a contraindication for surgery. Following successful surgery, there may be short-term worsening of psychiatric symptoms, in particular anxiety, but this is often followed by long-term improvement.27 Nevertheless, de novo psychiatric disorders, such as depression, anxiety or psychosis may develop in up to 26% of people after temporal lobe surgery. Careful post-operative psychiatric supervision is therefore important.27 Interestingly, new onset of dissociative (non-epileptic) seizures may develop in 4%–8% of people following surgery.28

Medication withdrawal

Patients who become seizure free following surgery may wish to consider subsequent medication withdrawal. However, the risk of seizure recurrence, the nature of prognostic factors, and the timing and rate of medication withdrawal remain unclear. Some studies suggest a high rate of recurrence, with failure to recapture seizure freedom on subsequently restarting medication,29 whereas others report that medication withdrawal is not clearly associated with an increased risk.30 In general, about one-third of people withdraw medication completely after surgery.4 31 However, with no clear data or guidelines, we prefer to maintain pre-operative medication for at least 12 months after surgery, before gradually reducing it to monotherapy during the second post-operative year, provided there is ongoing seizure freedom.

Recent developments

There is increasing interest in minimally invasive lesioning epilepsy surgery techniques. The philosophy of these treatments is that it may be possible to destroy or modulate a deep epilepsy focus without needing to sacrifice or damage potentially normally functioning surrounding brain tissue. They offer the possibility of stopping seizures yet with improved neurocognitive outcomes.

One such technique is thermal ablation, also called laser interstitial thermal therapy.32 The low-voltage laser is introduced via an optic fibre using MRI guidance and a coagulative necrotic lesion is created using MRI thermal maps in real time (figure 23). This is undertaken as a day case procedure and is becoming widely adopted for hippocampal sclerosis, hypothalamic hamartomas and cortical dysplasia. Compared with gamma-knife radiotherapy, there are no issues with long-term radiation risks and the region of ablation is better demarcated.

Figure 23

MRI-guided stereotactic laser amygdalohippocampectomy. Depiction of irreversible damage zone estimate at completion of ablation in the axial (A) and sagittal (D) planes. Post-ablation FLAIR image in the axial plane (B) and axial T1 image acquisition with gadolinium contrast enhancement (E). The post-operative FLAIR image with the estimated irreversible damage zone superimposed (C). Arrows indicate ablation damage zone estimate; arrowheads point to location of the implanted laser fibre assembly (reproduced from Wicks RT and colleagues by permission of Congress of Neurological Surgeons).32 FLAIR, fluid-attenuated inversion recovery.

In terms of outcome, real-time MR stereotactic laser amygdalohippocampectomy generally gives lower seizure free rates (53%–80% at 1–3 years) than open surgery in the small series completed to date. Hippocampal sclerosis on pre-operative imaging predicts a better outcome. Neurological complications, such as visual field defects, intracranial haemorrhage and cranial nerve defects occur at a rate comparable with open surgery, but neuropsychological decline (particularly of verbal memory) is less common.33 We await further assessment of the long-term seizure outcome and procedural risk from this and other lesioning techniques (such as high-frequency ultrasound ablation) and data on their potential use for more challenging surgical targets such as periventricular nodular heterotopia or corpus callosotomy.

Key points

  • All patients with refractory focal epilepsy should be considered for resective epilepsy surgery.

  • Pre-surgical evaluation—to characterise the epileptogenic zone and to prevent post-operative neurological, psychiatric and cognitive deficit—requires a multidisciplinary approach with a comprehensive investigative pathway.

  • Surgery carries a risk of permanent neurocognitive or psychological deficit but this needs to be balanced against the risks of ongoing seizure activity.

  • New techniques, such as laser ablation and responsive stimulation, aim to reduce epileptic activity while preserving neurocognitive function better than with standard resective surgery.


We are very grateful to Professor John Duncan for reviewing this manuscript.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.


  • Contributors FR-G drafted and revised the manuscript. AME and AM provided figures and revised the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests FR-G has received a speaker's honorarium from LivaNova. AME has received honoraria from Baxter, UCB and Integra, sponsorship to attend meetings from Leksell, Medtronic, Brainlab, Modus V and hospitality from Livanova. AM has received sponsorship to attend meetings from Medtronic and Modus V and received hospitality from Livanova. Research funding via Wellcome has Medtronic named as the preferred commercial partner.

  • Patient consent for publication Not required.

  • Provenance and peer review Commissioned. Externally peer reviewed by Khalid Hamandi, Cardiff, UK.

Other content recommended for you