Outcome of Epilepsy Surgery

Since the late 19th century, surgery has been accepted as a treatment for medically refractory epilepsy. Although a multicenter, randomized trial has not been performed, published data from individual centers and pooled data from the University of California Los Angeles (UCLA) Palm Desert Conference held in 1992 show that surgery for seizures results in seizure-free outcome in most patients with hippocampal sclerosis or a well-defined lesion.

Overview of seizure-based outcome classification systems

When looking at the efficacy of epilepsy surgery, a patient's postsurgical seizure frequency is compared with his or her preoperative seizure frequency. Usually, these assessments are performed at individual centers where the number of patients is small and patients serve as their own controls. In doing so, the assumption is made, based on the natural history of intractable epilepsy, that spontaneous remissions without surgery are infrequent. We now know from the work of Berg et al that prolonged remissions can occur before eventual intractability becomes evident.

Go to Epilepsy and Seizures, Neuroimaging in Epilepsy, Presurgical Evaluation of Medically Intractable Epilepsy, and Epilepsy Surgery for complete information on these topics.

Engel classification system

Numerous classification systems are used by epilepsy centers. The Engel classification system, devised in 1987, is the most commonly used scale. All seizure-outcome scales now in use contain subjective components, such as "worthwhile improvement" or "significant reduction." The Engel classification, for example, requires these subjective assessments to distinguish a class 3 outcome from a class 4 outcome, even though each center may have a different definition of worthwhile improvement. ^[1]

The Engel classification system is as follows:

• Class 1 - Free of disabling seizures (completely seizure free; nondisabling, simple partial seizures only; some disabling seizures, but free of disabling seizures for at least 2 years; generalized convulsion with antiepileptic drug withdrawal only)

• Class II - Rare disabling seizures (initially free of disabling seizures, but rare seizures now; rare disabling seizures since surgery; more than rare disabling seizures, but rare seizures for at least 2 years; nocturnal seizures only)

• Class III - Worthwhile improvement (worthwhile seizure reduction; prolonged seizure-free intervals amounting to more than half the follow-up period, but not less than 2 y)

• Class IV - No worthwhile improvement (significant seizure reduction; no appreciable change; seizures worse)

Class 1 includes patients with residual auras. Usually, auras do not bother the patient if they are infrequent. However, depending on the frequency and the nature of the auras (eg, intense fear), they can affect postoperative quality of life (QOL), even though they do not affect driving ability or independence.

In fact, Vickrey et al found that, of patients who undergo surgery, those with residual auras have less improvement in QOL outcomes than do those who do not have residual auras. ^[2] Furthermore, in some patients, residual auras can be associated with unrecognized loss of awareness.

Moreover, seizure-related disability is subjective. For example, for a patient who is not working and does not drive, 4 seizures a year may not be disabling unless they result in injury or inability to live independently, whereas in an individual who works and drives, as few as 1-2 seizures a year would be disabling.

ILAE classification system

In 2001, the International League Against Epilepsy (ILAE) proposed a new classification scheme for outcome with respect to epileptic seizures after surgery. The goal of this classification scheme was to provide a more objective measure of the number of seizures. However, this scale too becomes ambiguous after level 3; determining 50% reduction in seizures can be difficult to assess, since patients often do not maintain accurate seizure calendars. ^[3]

This classification system is as follows:

• Outcome classification 1 - Completely seizure free; no auras

• Outcome classification 2 - Only auras; no other seizures

• Outcome classification 3 - 1 to 3 seizure days per year; ± auras

• Outcome classification 4 - 4 seizure days per year to 50% reduction of baseline seizure days; ± auras

• Outcome classification 5 - Less than 50% reduction of baseline seizure days; ± auras

• Outcome classification 6 - More than 100% increase of baseline seizure days; ± auras

Other considerations in outcome classification

In addition to seizure freedom or reduction, there are other outcome measures, such as developmental and cognitive outcome, behavioral and psychosocial outcome, and improvement in health-related quality of life (HRQOL). In this regard, a seizure outcome scale should reflect dynamic changes in seizure outcome. It must differentiate between freedom from seizures and QOL. For example, a useful scale must be able to assess whether a patient who is not seizure free after surgery but has had prolonged periods of remission or a reduction of seizure frequency since surgery has had an improvement in QOL.

Vickrey et al demonstrated the variability in seizure classification systems in 1995. ^[4] The investigators compared 7 previously published seizure-based classification systems to an external standard, a self-reported HRQOL system specifically designed for patients with epilepsy. They found that the classification systems varied in their degree of correlation to this external standard.

Since the ultimate goal of epilepsy surgery is improvement in HRQL and functioning, seizure-outcome scales need to be objective and reflect dynamic changes in seizure outcome. This will help us to determine whether certain degrees of postoperative seizure control correlate with HRQOL measures.

Mesial temporal sclerosis is the most common indication for epilepsy surgery in adults, accounting for 75% of cases in adult surgical series.

In adults, the most common surgical procedure for temporal lobe epilepsy (TLE) involves the medial temporal lobe—either an amygdalohippocampectomy (resection of the amygdala, hippocampal head and body, and adjacent parahippocampal gyrus) or an anteromedial temporal resection that also includes resection of the anterior, inferior, and middle temporal gyri (3-4 cm from the temporal tip).

In children, cortical dysplasia and tumors are the most common etiologies, with neurocutaneous disorders (ie, tuberous sclerosis, Sturge Weber syndrome), hemispheric syndromes (Rasmussen syndrome, hemimegalencephaly), perinatal epileptic encephalopathies, hypothalamic hamartomas, and mesial temporal sclerosis being less common.

Respective surgery in children often involves resection of lesions, such as malformations of cortical development (ie, cortical dysplasia), hamartomas, developmental tumors, or lesions associated with perinatal, hypoxic, ischemic injury.

In a randomized, controlled, intent-to-treat trial comparing epilepsy surgery with medical treatment, Wiebe et al found that at the end of 1 year, 58% of patients in the surgical group were free from disabling seizures, and 10-15% had little or no improvement, compared with only 8% free from disabling seizures in the medical group.

The study also showed that QOL at 1 year was significantly improved in the surgical group, compared with that in the medical group. Morbidity was minimal in both groups. However, 1 patient died in the medical arm; no mortality occurred in the surgical group.

In a pediatric temporal resection series, 13-40% had mesial temporal sclerosis, most being adolescents. The seizure freedom rates were similar to those attained in adults, ranging from 58-78%. For pediatric temporal neocortical resections, 59-70% attained seizure freedom. Factors predicting a seizure-free outcome included (1) a focal lesion on magnetic resonance imaging (MRI) scans, (2) complete lesion resection, (3) localized seizures on EEG, and (4) the absence of preoperative, generalized tonic-clonic seizures.

Predictors of seizure-free outcome at 1-2 years

Although, as previously stated, evidence indicates that in most patients with hippocampal sclerosis, surgery for seizures results in a seizure-free outcome, the presence of hippocampal sclerosis is not an independent predictor for this result. Only 50% of patients with hippocampal sclerosis become seizure free if noninvasive data, such as interictal EEG, video-EEG recording, and neuropsychological testing, are not considered.

In a Mayo Clinic series, the factors predictive of freedom from seizures at 1 year after surgery were MRI-detectable, unilateral hippocampal formation atrophy and concordant interictal epileptiform discharges (IEDs). ^[5] The investigators found that 90% of patients with unilateral temporal interictal spikes, concordant ictal onsets, and unilateral hippocampal atrophy became seizure free after surgery. On the other hand, patients with bilaterally independent, temporal, interictal epileptiform abnormalities did not have as good a seizure-free outcome rate (ie, 70%).

In patients in the study with MRI scans showing hippocampal atrophy and IEDs discordant with ictal onset, only 60% had a seizure-free outcome. In patients with normal MRI findings, surgical outcome was similar, regardless of whether IEDs were concordant with ictal onsets.

In another study, which utilized 93 patients with mesial TLE who had anterior temporal resections, Jeong et al found that, on multivariate analysis, the only factors predictive of seizure-free outcome were age at surgery and the presence of hippocampal sclerosis.

Long-term outcome

The length of postsurgical follow-up is important in determining seizure-free outcome and HRQOL outcome. ^[6] For example, we need to know if epilepsy surgery produces a long-term outcome that is better than the natural history of the disease. Many studies have variable follow-up ranging from only 6 months to 2 years. During the 1990s, a few single center outcome studies were published that include reports of 5-year outcome, but some of these studies did not distinguish outcome for the different types of resective surgeries.

Rougier et al reported seizure outcome in 100 patients, with 66% of them seizure free at 1 year, and 51% of the patients seizure free at 5 years. ^[7]

Berkovic and colleagues found that 50% of mesial temporal sclerosis patients, 69% of foreign-tissue-lesion patients, and only 21% of nonlesional patients were seizure free for 5 continuous years. ^[8]

A study by Sperling et al indicated that seizure-free outcome at 2 years predicts long-term outcome. They found that 55% of seizure recurrences occurred within 6 months of surgery and that 93% of them occurred within 2 years after surgery. Five years after temporal lobectomy, 70% had an Engel class 1 outcome, and 11% had a greater than 80% reduction in seizures. ^[9]

Spencer reported the 2- to 10-year outcome in 226 patients who underwent epilepsy surgery. ^[10] They divided the seizure-free outcome into pathologic subgroups: glioma (70%), developmental (58%), vascular (60%), and mesial temporal sclerosis (67%).

A systematic review and meta-analysis by Engel et al that included 32 studies and 2,250 patients showed that 63% of patients remained seizure free after 2-5 years of follow-up. ^[11]

Similarly, Tellez-Zenteno and colleagues, in a meta-analysis of 40 studies that included 3,895 patients, showed that long-term follow-up of more than 5 years was 66%. ^[12]

In retrospective review of 143 patients with malformations of cortical development, sustained overall rates of seizure control were demonstrated in 72% at 2 years, 65% at 5 years, and 67% at 10 years. Complete resection of electrocorticographic and anatomic abnormalities was most predictive of seizure-free outcome. ^[13]

A study examining the long-term outcome of neurosurgical treatment on refractory focal epilepsy found that 52% of patients remained seizure free at 5 years and 47% remained seizure free at 10 years. Seizure recurrence was more likely in patients who had extratemporal resections than in those who had anterior temporal resections. No difference from anterior lobe resection was found in patients who had lesionectomies. ^[14]

Long-term outcomes are less well studied in pediatric patients than they are in adults. In a single-center study from the Montreal Neurologic Institute of 109 pediatric patients, Mittal and colleagues found that 86% of patients were seizure free or had a greater than 90% reduction in seizures with a median follow-up of 10.9 years. ^[15]

A long-term follow up study by Skirrow et al reported excellent long-term seizure control and favorable cognitive outcome, along with positive effects on brain development, when temporal lobe resection surgery is performed in childhood. ^{[16, 17, 18]}

In patients with TLE, the postsurgical seizure-free rate can only improve as more advanced imaging techniques become available. ^[19]

Epilepsy surgery outcome in TLE with normal MRI

In the current era, for an MRI scan to be considered normal, a dedicated epilepsy protocol MRI needs to be done. Many studies of TLE outcome with normal MRI were completed in the late 1980s and early 1990s, with seizure-free rates ranging from 18-63%. Today, many of these patients would have had an abnormal MRI, biasing outcome to be more favorable.

Holmes et al reported outcome in 17 patients with normal MRI and concordant noninvasive evaluation (ie, interictal and ictal electroencephalograms (EEGs)) without positron emission tomography (PET) or single-photon emission computerized tomography (SPECT) scanning, and found that 48% of the patients became seizure free and that 39% of the patients had a greater than 75% reduction in seizures after anterior temporal lobectomy (ATL). ^[20]

Sylaja et al also reported outcomes in patients with TLE with normal MRI and concordant EEG and found that 41% of the patients became seizure free and 29% of them had a greater than 75% reduction in seizures. Patients with unilateral anterior temporal interictal epileptiform discharges were more likely to have a seizure-free outcome. ^[21]

Areas for future investigation include whether lateralized WADA testing determines outcome in these patients or subtle MRI changes affect outcome. Furthermore, it is imperative to determine if long-term seizure outcome greater than 2 years remains favorable and to determine memory outcome in these patients, particularly those with dominant temporal lobe resections.

A retrospective study from the Mayo Clinic of 44 patients ^[22] with a nonlesional epilepsy protocol MRI identified preoperative factors associated with an Engel class 1 outcome (free from disabling seizures). These included (1) absence of contralateral temporal or extratemporal interictal epileptiform discharges, (2) subtraction ictal SPECT abnormality co-localized to the resection site, and (3) subtle MRI changes felt not to be significant at the time of the epilepsy surgery conference. Subtle MRI changes included increased T2 signal in the hippocampus without atrophy, abnormal enlargement of the amygdala, and subtle hippocampal atrophy without fluid-attenuated inversion recovery (FLAIR) signal.

A limitation of this study was that fewer than half of the patients had a fluorodeoxyglucose (FDG)-PET scan, which is known to predict seizure-free outcome.

In summary, unilateral temporal FDG-PET hypometabolism, subtraction ictal SPECT scanning, unilateral interictal spikes (anteromedial, not lateral), and localized EEG pattern can select patients with TLE with normal MRI who are likely to become seizure free with epilepsy surgery. However, 20-25% of patients with presumed TLE and normal MRI do not become seizure free or have a substantial reduction in seizure frequency. This latter group likely represents patients with (1) cryptogenic neocortical TLE, (2) temporal insular epilepsy, or (3) paradoxical TLE (extratemporal neocortical epilepsy with secondary propagation to the medial temporal lobe).

For this latter challenging group, prospective studies of larger numbers of patient using magnetic source imaging (MSI), FDG-PET scanning coregistered to MRI, and subtraction ictal SPECT scanning coregistered to MRI need to be done.

One noteworthy study of 9 neocortical TLE patients, by Knowlton and colleagues, ^[23] found that such multimodality imaging identified 78% of patients who became seizure free and 22% of those with rare disabling seizures.

A final word of caution for these cases is that in patients with dominant TLE and normal language and/or verbal memory function, surgery should only be pursued where seizures are significantly disabling, seizure-free outcome is estimated to be least 50%, and the patient is aware of the cognitive risks for temporal resection with a normal MRI.

In adult patients with extratemporal epilepsy and a well-circumscribed lesion, such as a developmental tumor (ganglioglioma, dysembryoplastic neuroepithelial tumor [DNET], or pleomorphic xanthoastrocytoma [PXA]) or a cavernous malformation, seizure-free outcome rates are in the range of 60-80%. Usually, seizure-free outcome is related to complete lesion resection.

Unfortunately, most studies do not separate outcome based on pathology and lobe of resection and also include temporal neocortical resections. Furthermore, most studies are single center, may not have long-term follow-up, have variability in presurgical evaluation and resective strategy, and have a relatively smaller numbers of cases.

Separating temporal neocortical resections is important, since these patients tend to have a more favorable outcome than extratemporal neocortical resections. In general studies that also included nonlesional cases, 36-76% of patients attained seizure freedom. Factors that predict a seizure-free outcome include the following:

• Presence of a discrete lesion on MRI

• Complete resection of the lesion

• Localized scalp EEG onset

• Concordant hypometabolism on PET scanning (with localized EEG or lesion on MRI)

• Longer duration of epilepsy

Usually, focal PET scan hypometabolism is more likely to predict outcome when there is a subtle, previously unrecognized lesion on MRI and is redundant when a well-defined lesion is seen, unless multiple regions of hypometabolism are observed.

A study of occipital lobe epilepsy by Aykut-Bingol et al at Yale found that the type of lesion can predict outcome. In the study, 85% of patients with tumors had an excellent/good outcome, while only 45% of patients with developmental lesions had an excellent/good outcome. Within the group with developmental lesions, cortical dysplasia was associated with a better outcome than heterotopia and hamartomas. ^[24]

Two separate single-center studies involving surgical treatment of lesional occipital lobe epilepsy from the University of Bonn and the Cleveland Clinic reported 69% and 81% seizure-free outcome (Engel Class I), respectively, at long-term follow-up (mean duration of 54 mo and 80 mo). ^{[25, 26]}

Binder et al reported on surgical treatment of lesional parietal lobe epilepsy, with 58% of patients becoming seizure free (mean follow-up of 45 months). ^[27]

Data from the study by Sisodiya of the Montreal Neurologic Institute showed that for patients with less well-circumscribed lesions, such as malformations of cortical development (MCDs), seizure-free outcome rate is approximately 47%. ^[28]

Kral et al reported 76% seizure free after resection of type 2a or 2b cortical dysplasia; however, this series included a few temporal cases (8 of 47). ^[29] The seizure freedom rate is likely higher in this latter study because type 2 cortical dysplasia is usually present on MRI and lesional cases typically have a better outcome, and because of the MCDs, focal cortical dysplasia usually has a better outcome.

A study by Janszky et al assessed predictors of outcome in lesional frontal lobe epilepsy and found that a generalized spike-and-wave EEG pattern adversely affected outcome. ^[30]

Mosewich et al analyzed predictors of outcome in frontal lobe epilepsy and found that the presence of a lesion, the absence of febrile convulsions, and postsurgical seizure control at 1 year were associated with an excellent outcome. ^[31] In a small series of 14 patients from the Mayo Clinic, more than 70% of patients with frontal lobe encephalomalacia became seizure free. ^[32]

Jeha et al reported seizure-free outcome in a patients with lesional and nonlesional frontal lobe epilepsy. The overall seizure-free outcome was 56% at 1 year, 45% at 3 years, and 30% at 5 years. Those patients with a circumscribed MRI abnormality had a 79% seizure-free rate at 5 years. Predictors of poor outcome included patients with a normal MRI scan, a lesion extending beyond the frontal lobe, generalized/nonlocalized EEG patterns, and incomplete surgical resection. ^[33]

In nonlesional extratemporal neocortical epilepsy, only 20-25% of patients have a seizure-free outcome, with an additional 25% of patients having worthwhile improvement. However, this outcome can be improved to 45% seizure free and 21% with rare disabling seizures with the use of functional imaging data (PET scanning, SPECT scanning, MSI) and scalp EEG. ^{[23, 34]}

Furthermore, certain scalp EEG patterns in nonlesional extratemporal epilepsy, such as focal high-frequency activity (FHFA) (>20 Hz), in conjunction with functional imaging, can predict seizure-free outcome, with 82% of patients becoming seizure free in a Mayo Clinic series. Only 21% of patient with other EEG patterns attained seizure freedom. ^[35]

In this regard, FHFA likely indicates that the ictal generator is close to the cortical surface making intracranial EEG more localizing than more distant or deeper ictal EEG generators, which typically produce less localizing propagated, ictal onset patterns.

Levesque et al defined dual pathology as a temporal or extratemporal lesion with coexisting hippocampal atrophy on MRI. ^[36] Typically, dual pathology is more commonly seen in children than in adults.

Li et al reported that hippocampal atrophy is present more frequently with certain types of lesions, such as MCDs, developmental tumors (ie, gangliogliomas, dysembryoplastic neuroepithelial tumors), and perinatal lesions (eg, porencephaly), than it is with cavernous malformations and low-grade gliomas. ^[37]

In patients who do have gliomas or cavernous malformations, hippocampal atrophy is present more frequently when the lesion is medial (ie, medial to the collateral sulcus) than when it is lateral. In patients with temporal lobe lesions and coexisting hippocampal atrophy, freedom from seizures is not obtained unless the lesion and the atrophic hippocampus are included in the resection.

Li et al found that seizure freedom was attained in 73% of patients with a lesionectomy and a medial temporal resection (resection of hippocampus, amygdala, and parahippocampal gyrus), whereas seizure-free outcome was only 20% in patients with medial temporal resection alone and only 12.5% with lesionectomy alone. ^[38]

In patients with nontumoral occipitotemporal epilepsy, Aykut-Bingol et al found that patients who had only a hippocampal resection did poorly, whereas most patients with a combined occipital and temporal resection and some patients with an occipital lobe resection had a favorable surgical outcome. ^[39] Malformations of cortical developmental were the most common pathologic substrate in their study.

In patients with an occipital lobe lesion with semiologic features and/or EEG features of medial temporal lobe propagation, a lesionectomy without hippocampectomy is reasonable. In some cases in which noninvasive data are equivocal, invasive EEG is required to determine whether independent ictal onsets occur in the temporal and the occipital lobe. In patients with occipital lobe epilepsy and ipsilateral hippocampal atrophy, a combined resection is likely to yield a greater chance for a seizure-free outcome.

Studies have indicated that surgery in dual pathology has similar seizure-free rates if the hippocampus and the additional lesion or cortical dysplasia are both completely resected. However, most of these series were small and did not include long-term follow-up.

Surgical outcome in children with focal resections is similar to that in adults, with seizure-free rates in the range of 75-80%, particularly in patients with hippocampal sclerosis or tumors. However, the pathologic substrate of surgically treated epilepsy is different in children than in adults. For example, hippocampal sclerosis, the most common pathologic substrate in medically intractable TLE of adulthood, usually does not result in refractory seizures until adolescence or early adulthood.

MCD and low-grade tumors, on the other hand, are the most frequent causes of intractable epilepsy in childhood. For example, MCDs, such as focal cortical dysplasia (FCD), ^[40] are observed in greater than 50% of surgically treated pediatric patients, whereas they are observed in only 20% of surgically treated adult patients.

In children, as in adults, outcomes after resection of MCDs are less favorable ^[41] than they are after resection of well-circumscribed lesions, such as low-grade gliomas and cavernous malformations, with only 50% of pediatric patients attaining freedom from seizures after MCD resection.

In the UCLA series of pediatric cases, 56% of extratemporal unilobar epilepsy cases were seizure free at 2-5 years. ^[42] Similarly, in the Miami Children's Hospital series of patients with FCD, 55% were seizure free at 2 year follow-up. ^[43] Late seizure recurrence or worsening of seizures took place in 17% of patients on long-term follow-up, occurring more frequently in patients with FCD type II.

On the other hand, 8% of the patients, often those with minor MCDs or FCD type I, improved or became seizure free after the second postoperative year. The most important predictor of outcome in these series was completeness of resection.

Patients with multilobar resections have a less favorable outcome, with Wyllie et al, ^[44] Mathern et al, ^[45] and Hemb et al ^[42] reporting that approximately 52-55% patients became seizure-free.

In summary, outcomes in extratemporal lobe resections are similar in children and adults. Outcomes are better with well-circumscribed lesions, such as developmental tumors and cavernous malformations, than they are with FCD, and outcome correlates with completeness of lesion resection and unilobar localization.

Patients without an MRI-identified lesion and those requiring a multilobar resection have a less favorable outcome. For further review of this topic, the reader is referred to an excellent critical review by Lerner and the UCLA group that outlines the age of onset; EEG, MRI, and functional imaging findings; and surgical outcome in, as well as future direction for study of, FCD. ^[46]

Hemispherectomy/functional hemispherectomy

Usually, this procedure is reserved for infants and children with catastrophic epilepsy, developmental regression, and a unilateral useless hand. The usually pathologic substrates include hemimegalencephaly, perinatal infarction, diffuse MCDs, Sturge-Weber syndrome, and Rasmussen encephalitis.

Multiple single-center series from the Cleveland Clinic, Johns Hopkins Hospital, and Miami Children's Hospital reported seizure-free outcome rates in the range of 53-67%. ^{[47, 48, 43, 49]} Overall, their seizure surgery outcomes were better in patients with non-MCD pathologies (ie, perinatal infarction, Sturge-Weber syndrome) than in those with MCD.

Similar outcomes were achieved by Devlin and colleagues of the London group, again showing better outcomes with acquired pathology than developmental pathologies (ie, MCDs). ^[50]

Researchers in British Columbia reported 79% of patients as seizure free after a median 7-year postoperative follow-up. ^[51]

In the large UCLA series of 141 cases, the overall rate of seizure-free outcome with hemispherectomy was 83% for cases after 1997 and 66% for cases from 1986-1997. ^[42] No differences were reported in 2-year outcomes between the groups with MCD and non-MCD pathologies. However, in a follow-up interval of 2-5 years, 25% of patients with cortical dysplasia (hemispherectomy, multilobar, and lobar cases) had recurrent seizures. ^[42]

Corpus callosotomy

Corpus callosotomy is a palliative disconnection procedure in which the corpus callosum is sectioned to prevent the interhemispheric propagation of seizures and generalization of seizures that results in tonic/atonic seizures and generalized tonic-clonic seizures. The procedure involves disconnection of the anterior two thirds of the corpus callosum or complete callosal resection.

This form of surgical treatment is recommended for children and adults with symptomatic generalized epilepsies, such as Lennox-Gastaut syndrome, who have disabling atonic and/or tonic seizures. These children usually have clusters of tonic seizures, which result in sudden falls, causing head injury. These seizures severely limit the daily activities of these developmentally disabled patients.

Corpus callosotomy is a palliative procedure to limit or modify tonic/atonic seizures; it rarely makes patients seizure free. The operation results in an 80% average reduction in tonic/atonic seizures resulting in falls, a 50% reduction in generalized tonic and tonic-clonic seizures, and 50% atypical absence seizures (although this seizure type is often difficult to quantify). The seizures still occur as partial seizures, but they do not result in falls. Overall, success rates are similar between children and adults and the effects are usually sustained long term.

After corpus callosotomy, most patients have significant improvement in cognitive function, activities of daily living, and behavior. Outcome is better with a complete callosotomy than with an anterior two-thirds callosotomy. However, complications are more common with complete callosal resection, including disconnection syndrome (mutism, hemiataxia, alexia).

Typically, an anterior two-thirds callosotomy is performed, with complete callosotomy reserved as a second procedure if there is inadequate relief of drop seizures after the first procedure.

With the introduction of newer drugs (eg, felbamate [1993], lamotrigine [1995], topiramate [1997], zonisamide [2000], rufinamide [2009], and clobazam [2011]) and the vagal nerve stimulator, the number of patients requiring corpus callosotomy has been reduced.

Memory outcome s

Since the hippocampus is important in memory and learning, anterior temporal resections have a potential risk of producing a memory deficit. Bilateral injury to the hippocampi is known to produce long-lasting, profound, anterograde amnesia. Unilateral ATL does not produce a severe loss of memory unless the contralateral hippocampus is unable to sustain memory. In this regard, the intracarotid amobarbital test is essential prior to surgery to evaluate the ability of the contralateral hippocampus to sustain memory after unilateral ATL.

Material-specific memory impairment may occur after ATL. Significant impairment of verbal memory is a potential complication of dominant ATL. Individuals with average or above-average memory function prior to surgery are at greatest risk of postoperative memory decline.

After right ATL, most patients have an overall improvement or no change in memory function. Certain studies suggest, however, that some patients with high preoperative memory scores who undergo right ATL may experience a modest decline in memory after surgery.

Long-term studies with more than 5 years follow-up suggest that the memory deficit appears early after surgery and that the rate of decline stabilizes after 1-2 years.

Language outcomes

Dominant-hemisphere ATL has a selective risk of postoperative decline in visual confrontation naming. The standard ATL includes resection of the middle temporal gyrus (MTG) and the inferior temporal gyrus (ITG) to approximately 4.5 cm and the superior temporal gyrus (STG) to 2 cm.

The classic studies of Ojemann et al showed that few language sites are observed in the anterior 5 cm of the MTG and ITG; however, language sites may be observed in the first 2 cm of the STG. Numerous studies have demonstrated that patients with a high preoperative score on the Boston Naming Test (BNT) and a later age of onset (>10 y) are at greatest risk of a postoperative decline in naming.

Many centers perform an anteromedial temporal type resection (ie, Spencer type), which, unlike standard ATL, spares the STG and involves a less-extensive lateral resection of the MTG and the ITG. A few centers perform intraoperative language mapping (IOLM) with tailored resection.

However, studies have shown that language outcomes at 1 year in patients treated with anteromedial-type resection are no different than the outcomes of patients who underwent tailored resection with IOLM.

Patients with a late age of onset and high BNT scores should be counseled that they are likely to have a significant reduction in their postoperative naming ability after dominant ATL.

Psychiatric outcomes

Psychiatric disorders, particularly depression and anxiety, are seen in up to 50% of patients with epilepsy. Furthermore, patients with psychiatric comorbidity with intractable epilepsy have worse HRQOL than those patients without. In fact, in patients with medically intractable epilepsy, depression predicts worse QOL, whereas seizure frequency has no predictive value.

In the largest prospective study to date, Devinsky and colleagues reported on 360 patients, 89% of whom had temporal lobectomies, and found moderate and severe levels of depression and anxiety in 22% and 24% of presurgical patients, respectively. However, these rates declined over 2 years to 11% and 12%, respectively. At 24 months, moderate to severe depression was seen in only 8% of those patients who were seizure free. ^[52]

Patients with focal epilepsy who underwent surgical treatment for intractable seizures were predisposed to greater depression after surgery. The psychosocial outcome after epilepsy surgery appears to be intrinsically linked to a change in self and to difficulties patients may have in adjusting to freedom from seizures. ^[53]

Patients with epilepsy hope to have an improvement in their HRQOL, employment, education, and social activities. Improvement in these domains may be more difficult to achieve the longer the epilepsy has remained uncontrolled.

Vickrey et al showed that the degree of seizure control strongly correlates with scores on the Epilepsy Surgery Inventory-55 (ESI-55), a self-reported measure of quality of life. ^[4] Vickrey et al studied quality-of-life outcomes in 248 patients who had a diagnostic evaluation for epilepsy surgery. They found that patients who underwent surgery scored higher on 5 of 11 scales at follow-up than did patients who did not undergo surgery for epilepsy. Vickrey et al found an improvement in HRQOL in 170 patients who were seizure free after surgery. Patients with residual auras had improved HRQOL, but less so than those who were completely seizure free had. ^[54]

A prospective surgical series of 396 patients in which the Quality of Life in Epilepsy Inventory-89 (QOLIE-89) was used before surgery and up to 5 years after surgery, the largest gain in HRQOL occurred immediately after surgery in all patients, with further improvements in the seizure-free group, and stabilized at 2 years. ^[55] In the group with the most favorable seizure-free outcome, HRQOL was comparable to that seen in the general population.

Similarly, Helmstaeder et al reported severely impaired QOL scores in 33% of medically treated patients, as compared with 10% of surgically treated patients, with intractable epilepsy. Furthermore, severely impaired QOL was seen in 3% of surgically treated patients who were seizure free, as compared with 31% of patients who had continued seizures.

Langfitt et al found that patients in remission after ATL at 2 and 5 years had improved HRQOL regardless of memory outcome. Among those patients not in remission at 2 and 5 years, HRQOL remained stable when memory did not decline, but HRQOL declined when memory declined. ^[56]

Therefore, patients who are not seizure free with epilepsy surgery and have memory decline have reduced HRQOL, a factor that needs to be considered in patients with dominant TLE with a baseline relatively preserved verbal memory and a reduced chance of attaining seizure freedom with epilepsy surgery.

In patients with unilateral medial TLE or a well-circumscribed temporal or extratemporal lesion and concordant interictal and ictal EEG data, seizure-free outcome rates range from 60-90%, with temporal lobe lesion cases having a more favorable outcome.

In patients with MCDs and interictal/ictal EEG data that are concordant with the MRI lesion, rate of seizure-free outcome is still favorable but is reduced to 50%.

In patients with nonlesional neocortical epilepsy, on the other hand, the seizure-free outcome rate is only 20%, with an additional 25% having a reduction in seizure frequency of at least 80%.

Fortunately, with the development and refinement of imaging techniques such as high-resolution structural MRI, FDG-PET/MRI coregistration, ictal SPECT with statistical parametric mapping, MSI, and diffusion tensor imaging, a greater number of patients without a discrete lesion can be identified to improve seizure-free outcome to 50-60%.

In this author's experience, when there are multiple concordant studies, a discrete localization within a lobe, scalp focal ictal paroxysmal fast activity, and noncontiguity with eloquent cortex, outcome in nonlesional cases is optimized. Furthermore, functional imaging studies on novel structural imaging, such as susceptibility weighted imaging and diffusion tensor imaging, can increase subtle or occult lesion conspicuity, thereby improving outcome until it is similar to that in patients with more obvious lesions.

Factors that are likely to predict a favorable, seizure-free outcome include (1) absent or infrequent secondarily generalized convulsions, (2) the presence of a circumscribed lesion, (3) the absence of diffuse pathology, (4) complete lesion resection, and (5) type of pathology. Furthermore, seizure-free outcome at 2 years predicts long-term outcome at 5 and 10 years.

The most common sequela of dominant ATL, postoperative decline in naming, is related to later age of onset of epilepsy and high preoperative BNT score. Patients with early age of onset and low preoperative BNT score are at low risk of a postoperative decline in naming ability.

The ultimate goal of epilepsy surgery is to improve QOL and social and occupational function. Several studies in this regard have shown that HRQOL, occupational outcome, and psychiatric outcome are improved; epilepsy-related mortality is reduced; and direct and indirect health care costs are greatly reduced.