- Top of page
Summary: Purpose: To review systematically the available evidence with regard to the current status of epilepsy surgery in the management of patients with epilepsy.
Methods: A careful search of published literature, including Medline, published reviews, chapters, and cross-references thereof.
Results: With medical treatment of epilepsy being unsuccessful in many cases, the importance of surgical approaches cannot be underscored. Early surgery is the treatment of choice for patients with clear-cut mesial temporal sclerosis and results in significant clinical improvement in up to 80% of cases, provided the EEG, neuropsychological, and neuropsychiatric results are in concordance with this approach. In patients with poorly defined, widespread, or dual pathology, however, invasive recordings may be necessary, and while this is performed in major centres, the outcome is rather more variable in this group. Improved surgical techniques, and the use of stereotactic approaches and image guidance procedures, have resulted in surgical resections becoming more selective. With isolated structural lesions such as dysembryoplastic tumours, low-grade astrocytomas, or focal vascular abnormalities, total macroscopic and radiological evidence of lesional excision is associated with excellent seizure-free outcome. The first randomised controlled trial of epilepsy surgery has demonstrated clearly the efficacy of these techniques, and the risk of complications.
Discussion: Increasing sophistication of noninvasive presurgical evaluation enables surgical candidates to be identified at an earlier stage and presents a realistic alternative to medical treatment in many cases. The introduction of minimally invasive techniques has had a significant impact on surgical practice and its associated morbidity. The future of epilepsy surgery lies with continued basic science research and its application to clinical medicine.
Since the first procedures were carried out for the surgical treatment of epilepsy, it has taken almost 120 years for a randomised controlled study to be performed and to demonstrate clearly the efficacy of surgery when compared with medical treatment (1). The study, published in the New England Journal of Medicine, randomly assigned 80 patients with seizures suspected of being of temporal lobe origin to either medical or surgical treatment. Ninety percent of those patients allocated to surgical treatment underwent a resection, and 58% of those became free of seizures impairing awareness in comparison with 8% in the medical group. It was previously merely accepted that surgery offered a potential cure for those who were refractory to medical treatment, with success rates for temporal resections varying from 53 to 84%(2), and it is only now, with improved diagnostic techniques and understanding of the consequences and risks of continued seizures, that surgery is being considered a real alternative to medical management earlier during the course of the patient's treatment.
The recently published report from the National Institute for Clinical Excellence in the United Kingdom (3) not only reinforces the risks of uncontrolled epilepsy but also throws light on how badly epilepsy can be managed medically, as a result of a number of prevailing circumstances (4). Despite the increasing number of anticonvulsant drugs (AEDs) available and their proven efficacy in the majority of patients with epilepsy, there remain a significant percentage of patients who will not come under control with either first- or second-line AEDs (20–30%) and in whom this pattern of intractability may be identified early (5,6) in the management pathway, implying that many of the physical and psychological complications of epilepsy might be avoided. The effects of AEDs on cognition are difficult to evaluate (7,8) but at an anecdotal level, many patients and parents feel that the effects can greatly prejudice education and behaviour. At a cellular level, the overexpression of drug-resistance proteins in resected specimens of patients with drug-resistant epilepsy asks the question as to whether, in such cases, medical treatment is bound to fail (9). Imaging studies also showed that hippocampal volume loss in patients with refractory epilepsy can be progressive and at a rate higher than that of the normal population (10,11).
The selection of patients for surgical treatment depends on accurate presurgical diagnosis and identification of the underlying pathophysiologic substrate of the seizures. The general principle still prevails that resective surgery is curative, whereas functional neurosurgical techniques, although potentially offering the prospect of cure, are usually palliative. Mesial temporal sclerosis currently represents the commonest pathologic entity underlying uncontrolled epilepsy, and the association between the pathologic finding of Ammon's horn sclerosis and temporal lobe seizures was first described by Sommer in 1880 (12). The mechanism or mechanisms underlying this pathologic process remain uncertain, although both physiologic and genetic causes have been suggested. The prolonged febrile convulsion is clearly associated with the subsequent findings of hippocampal sclerosis and temporal lobe seizures, but as there is also a familial incidence of febrile convulsions, it is unclear whether single or multiple factors contribute (13). Penfield and Falconer (14) described the efficacy of temporal lobe resection based either on the intraoperative electrocorticograph or alternatively on an anatomic basis, stressing the significance of the findings in the pathologic specimen and, in addition, the importance of positive radiologic findings in the preoperative investigations. The application of magnetic resonance imaging (MRI) to the head in 1980 (15) and the subsequent recognition of mesial temporal sclerosis on MRI scan led to a dramatic increase in the number of surgical candidates being identified (16). After the initial description of reduced hippocampal size and increased T2 signal, a number of surgical series described the relation between the imaging findings and the underlying pathology (17,18). Neuronal loss and gliosis may have quite separate pathologic significance and, as a result of this, it has been possible to identify patients in whom one, but not both, of the radiologic features are present (19). Pathologic analysis of resection specimens when correlated to surgical outcome demonstrated the adverse effect of dysplastic neurones on outcome. Functional imaging techniques such as single-photon emission computed tomography (SPECT) (20) or positron emission tomography (PET) may give localising information about ictal cerebral blood flow, glucose metabolism, or benzodiazepine (BZD) receptors that implicate a more widespread physiologic abnormality, which may exceed the limits of the anatomic abnormality. In addition, quantitative techniques for the evaluation of the grey–white interface or cerebral convolutions have been used to identify malformations of cortical development; techniques such as diffusion tensor–weighted imaging offer potential for the future, as when the extent of the abnormality can be defined, then the outcome can be optimised.
Now that the pathologic substrate can be so clearly demonstrated and understood preoperatively, we can predict postoperative outcome not only in terms of potential freedom from seizures (21,22) but also in terms of potential neuropsychological deficit (23). Although anatomic imaging has assumed a prominent role in presurgical evaluation, it should not be carried out in isolation without appropriate neurophysiologic, neuropsychological, and neuropsychiatric assessments (24). The MRI data should be taken in the context of the EEG, psychometric strengths and deficits, and psychiatric status of the patient, as only then can the patient's candidacy for epilepsy surgery be properly determined. When clear-cut hippocampal sclerosis occurs in isolation, and other investigations are congruent, then the chances of being free of seizures will approach 80%. The presence of dual pathology, more widespread psychological deficit, or personality disorders may lead to poor outcome in relation to both seizures and psychosocial outcome, and this factor may result in surgery either being declined or deferred. By contrast, if a patient is an excellent surgical candidate, then presurgical evaluation and early surgery should be considered. The potential risks and benefits of surgery in each patient should be the subject of a multidisciplinary discussion. Thereafter the patient, the family, and if appropriate, their carers, should be counselled and consent to the operation and its potential consequences (25).
In the surgical management of mesial temporal sclerosis, the contribution of ictal EEG is currently being questioned (26,27) and where the interictal EEG, MRI, and neuropsychological and psychiatric data are congruent, it is proposed that surgery should take place without ictal recordings (28). The presence of bitemporal interictal spikes, widespread neuropsychologic deficits, or psychiatric disease is likely to remain a strong indicator for ictal EEG recording, but should the policy of reducing the need for video telemetry be adopted, it would have wide-reaching implications because of the reduced cost of presurgical evaluation. This would have far-reaching consequences, as the introduction of MRI obviated the need for invasive recording, and may result in a much more widespread application of epilepsy surgery. There is a continuing need for invasive recording to be carried out, but its use is generally restricted to complex diagnostic issues and pathologies and to specialist centres (29). When invasive recording is carried out, the technique used will depend on the nature of the pathology and the questions asked. A clear thesis for the origin of the epilepsy and its relation to any demonstrated pathology and eloquent areas is essential before the implantation takes place. Recordings from intraparenchymal depth electrodes provide excellent neurophysiologic information but sample a very restricted area. In contrast, subdural grids and strips cover a wider and more superficial area and are therefore less localising, but offer the opportunity of cortical localisation by means of stimulation. In many institutions, stereotactically implanted depth electrodes are now reserved for discriminating between seizures of mesial temporal lobe onset when the laterality is not clear.
The operative and anaesthetic techniques used in the surgical treatment of epilepsy have evolved in parallel to other branches of neurosurgery. Sir Victor Horsley, in his Linacre Lecture in 1909 entitled “The Anatomical Construction of the Cortex Cerebri,” described the method by which a subpial dissection of a gyrus may be achieved with preservation of the blood supply to adjacent gyri, the essential surgical technique for the practice of epilepsy surgery. Before field magnification with the operating microscope, Penfield (30) and Falconer (31) described specific approaches to the temporal lobe, and since the introduction of the microscope, the techniques have been refined and made less invasive. Selective surgical techniques targeted toward the mesial temporal structures began with the anatomic approaches of Niemeyer (32) and Yasargil (33) and have been further developed by Olivier (34) and others after the introduction of image guidance. The neurologic and neuropsychological effects of large resections are inevitably balanced against the lower incidence of seizure-free status with more limited resections (35). Transsylvian or image-guided techniques potentially offer lower risk of neuropsychological deficit. Epilepsy surgery also has its own procedures such as hemispherectomy, corpus callosum section, and multiple subpial transection (36). Hemispherectomy began as a procedure for glioma in the 1920s (37) but was applied to epilepsy by Mckenzie (38) and popularised by Krynauw (39). After the recognition of superficial haemosiderosis and its consequences (40), the anatomic procedure modified by Rasmussen (41) has since been refined by Delalande, Villemure (42,43), and others. Corpus callosotomy (44) and subpial transection (45) are both functional procedures, the former being discovered serendipitously, and the later, a technique based on an hypothesis applied after extensive animal investigation. The criteria for considering either procedure appear unclear, and the reported outcomes are variable.
Frame-based stereotactic techniques have been extensively used since 1953 (46), and since the introduction of image guidance, it has been used extensively to apply a minimally invasive approach to electrode implantation of both depth and subdural electrodes, as well as to localise lesions and facilitate anatomic resections. Image guidance has the additional advantage of teaching and assisting those who are unfamiliar with normal or abnormal anatomy intraoperatively, thus increasing the safety of the minimally invasive approach. The extent of any surgical resection should relate to the nature of the underlying pathology. In the case of isolated structural lesions such as dysembryoplastic tumours (47), low-grade astrocytomas (48), or focal vascular abnormalities (49), total macroscopic and radiologic evidence of lesional excision is associated with excellent seizure-free outcome. Application of image guidance to these cases greatly facilitates surgical excision. Fusion of anatomic and functional images on the image-guidance workstation also may introduce further information to the operating room, allowing localisation of functionally eloquent and functionally pathologic tissue. Most recently we have seen the introduction of intraoperative MRI to epilepsy surgery, which is used not only to plan the surgical route but also to ensure adequate resection of identified pathology (50,51). This technique represents a further refinement of the surgical technique, but the benefits of this approach in clinical outcome and financial terms have yet to be proven.
No surgical procedure can be carried out without risk of adverse events, and the New England study revealed that visual field defects occurred in 55%; in one patient, a small thalamic infarct developed, and two had a significant decline in verbal memory (1). In this study, an extensive temporal resection was performed, which we know has bearing on adverse events, but nonetheless, these adverse events must be compared with the single case of sudden death (SUDEP) that occurred in the medical group. The risks of chronic epilepsy should therefore be balanced against the risk of surgical adverse events and the patients counselled accordingly. Information from the presurgical investigation also allows prediction of neuropsychological deficits after temporal lobe surgery and may obviate the need for the sodium amytal test (52). The relation between reduced left hippocampal volume and verbal memory quotient can be accurately used to predict postoperative memory deficits (24,53,54).
The International League Against Epilepsy working party in surgery suggested a two-tiered structure for the surgical management of epilepsy (55), with a basic centre providing a service to adults with more straightforward surgically remediable epilepsy and an academic, reference centre, offering facilities for invasive recording and the treatment of children and basic scientific and clinical research. The implication of this is that the development of noninvasive protocols centrally will allow a wider application of the surgical approach to epilepsy and thus benefit a greater number of patients. Academic institutions must explore the benefit of potentially expensive, novel investigation techniques to evaluate developments in technology such as spoke-activated MRI or spike localisation with magnetoencephalography (MEG) (56). Video telemetry, electrocorticography, stereotactic facilities, MEG, and intraoperative MRI are all expensive and may, for the most straightforward of surgical cases, have little impact on outcome. It is essential that protocols for treatment be developed that minimise the cost of presurgical workup, while giving a high-percentage chance of surgical success (57). The surgical procedures involved must be ones that can be most widely and safely applied, to the benefit of the greatest number of patients.
In the United Kingdom, the potential pool of patients who could benefit from surgery has been assessed as being in the region of 4,500, and yet fewer than 600 operations are performed each year, and 27% of these are functional and therefore palliative (58). Despite the proven efficacy and potential of surgical treatment for epilepsy, it remains unheard of by many patients and clinicians, and a recent primary care article on the treatment of epilepsy failed to mention surgery at all (59). The experience of epilepsy surgery before MRI diagnosis and modern surgical and anaesthetic techniques was chequered, giving rise to resistance among clinicians to refer for treatment (60). Surgical series frequently have a short follow-up, but there is now increasing evidence to show maintained long-term seizure-free outcome (61–63) as well as quality-of-life improvements (64); indeed quality-of-life improvements may only emerge with time (65).
Earlier diagnosis and investigation of patients with epilepsy inevitably results in surgical candidates being identified within the paediatric age group. The pathology encountered in the paediatric epilepsy surgery population is very similar to that of the adult, but with a shift toward disorders of cortical development. Our understanding of the normal patterns of cortical migration is being continually updated, and the traditional concept of radial migration modified and added to, with the recognition of tangential migration (66,67).
Abnormalities of the different processes involved in cortical development can be correlated to the patterns of abnormality encountered clinically, and a greater understanding of the developmental processes may result in refinements of the surgical approach adopted. It has been shown that the neural circuits within disordered cortex are aberrant and that the pathophysiologic abnormality may be more widespread than the apparent anatomic abnormality. This may account for the improved outcome from wider resections involving both anatomically and neurophysiologically abnormal tissue and should be taken into account when planning such resections or disconnections. The occurrence of both catastrophic epilepsy of infancy and encephalopathic changes in other severe epilepsy syndromes of childhood means that surgery is often urgently required. Because of the nature of the developing nervous system, continuing seizures can have a profound effect on cognitive and motor development, as well as on behaviour, and the multidisciplinary presurgical assessment must include these areas. The surgical techniques used in children vary little from those in adults, but a close working relation with anaesthesia and meticulous technique are desirable.
In conclusion, despite the ever-increasing numbers of drugs available, a proportion of patients remain resistant to medical treatment. The increasing sophistication of noninvasive presurgical evaluation enables surgical candidates to be identified at an earlier stage and presents a realistic alternative to medical treatment is some cases. This approach affects the many patients with epilepsy in both the developed and developing countries. The introduction of minimally invasive techniques has had a significant impact on current surgical practice and its associated morbidity. The future of epilepsy surgery lies with continued basic science research and its application to clinical medicine.