The evolution of epilepsy surgery
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/neuroindia.NI_1028_16
Source of Support: None, Conflict of Interest: None
This review traces the evolution of epilepsy surgery from its early beginnings in the 20th century with the development of neurophysiology, and later the identification of pathology in surgical specimens, to the tremendous boost given by direct brain imaging in the late 20th century. This resulted in the sophisticated methods of presurgical investigation, surgical techniques, and postsurgery care available from the millennium. In parallel, functional surgery, which modifies the nervous system's behaviour, available throughout, has attained a greater place by the use of stimulation.
Keywords: Epilepsy, history, magnetic resonance imaging, stimulation, surgery
The evolution of epilepsy surgery depended upon both technological advances and understanding the pathophysiological basis of epilepsy. After the early period, when reliance was upon clinical data alone, there was a relatively stable period of using neurophysiological data and indirect neuroimaging. However, when pathology, discovered between 1930 and 1990, was combined with the results of direct imaging from computed tomography (CT) and magnetic resonance imaging (MRI), the two streams of surgery, which had always been present, were one which relied upon the structural lesion and another which sought to change the function of the brain. The first gave rise to operations resecting an identified lesion; the second spawned various methods of affecting function by destroying or isolating areas of the brain, and recently, of stimulating the brain.
The evolution of epilepsy surgery can be divided into four phases:
There is evidence that trepanation, which was used for various neurological conditions, was carried out in all Neolithic communities, including India, 4000 years ago. Thereafter the scene shifted to the Western hemisphere, where in the late 19th century, a number of surgeons in North America and Europe began operating on the basis of a clinical incident. For example, Macewen in Glasgow and Horsley in London operated on patients using the localization deduced from the work of Ferrier. Around the same time, that is the early years of the 20th century, in Germany, Fedor Krause was operating under similar constraints but added the use of monopolar Faradic stimulation to stimulate the cortex. He reported on 400 cases operated for epilepsy during his career. Otfried Foerster operated in the same manner and influenced Penfield; together they published a paper describing operations on traumatic scars of the cortex associated with their experimental work on such cicatrices. Although the paper concentrates on the mechanism of epileptogenesis in these cases, it also illustrates the surgical methodology used at that time. The final development in this phase was the ability to record the electrical activity of the human brain first described by Berger. This permitted recording from the scalp and subsequently directly from the cerebral cortex. Although occasional recordings were described, including those by Adrian and Matthews in 1934, the systematic exploration of cortical recording was made by Penfield and Jasper in Montreal. They described the technique in 1935 and summarized the use of these techniques in 1941. During this time, they also made considerable contributions to the localization of cerebral function because many of their procedures were carried out in cooperative adults under local anaesthesia. The volume and variety of their work also encouraged ground breaking research in neuropsychology including the development of the Wada test for speech and memory localization. In a summary of their experience, there is a chapter devoted to pathology, which described cerebral tumours in general and atrophic lesions in considerable detail. The development of another technique for examining the neurophysiology of epilepsy, which involves exploration of the brain with depth electrodes, was emerging in France. In the late 1940s and early 1950s, interest in epilepsy surgery emerged in a number of centres including Los Angeles, Yale, London, and India, where Professor B. Ramamurthi [Figure 1] began stereotactic procedures.
In the next phase, it was realized that localized pathology was important in the genesis of epilepsy and in obtaining relief by resective surgery. The majority of drug-resistant epilepsy cases, particularly at that time, were of temporal lobe epilepsy, and the pathology in 50% of such cases was hippocampal sclerosis. This lesion was first described in the late 19th century notably by Sommer  and Bratz. However, it was Murray Falconer's technique of en bloc resection of the temporal lobe that made detailed pathology of the temporal lobe available for analysis. At the same time, a study of temporal lobe patients by Margerison and Corsellis  showed the diversity of the distribution of hippocampal sclerosis, both between and within the hippocampi, a matter which became important later. Even with modern investigatory techniques, the causes of hippocampal sclerosis remain complex; however, the theory of incisural sclerosis favoured by the Montreal Neurological Institute has now been abandoned. The Montreal group, describing the epileptogenic lesions seen in their practice, divide them into expanding lesions, comprising various kinds of tumours, abscesses, haematomas, and atrophic lesions. Over the course of the next 30 years and before the availability of direct brain imaging, a number of other pathologies were emerging including malformations of cortical development, first described in surgical specimens as cortical dysplasia by Falconer and Taylor [Figure 2], and Rasmussen's encephalitis first described in 1958.
Prior to the early 1970s, the only methods of demonstrating structural abnormalities in the brain were indirect, unpleasant, and potentially dangerous neuroradiological methods of angiography and air studies, together with the keen eye of the neuroradiologist to observe the presence of a faint calcification. There were methods of detecting functional abnormalities in the brain using radionuclides. In the positron emission tomography (PET) scan, it was possible to detect the abnormalities of glucose metabolism using a radioactive isotope of fluorine substituting it for a glucose atom. In addition, it was possible to use single-photon emission computed tomography (SPECT) to demonstrate the epileptic focus, as was demonstrated by Berkovitch and colleagues simultaneously recording ictal SPECT studies and intracerebral electrode recordings. The first technique of imaging the brain directly was CT scanning, first described by Hounsfield. Approximately 10 years later, the technique of MRI emerged as an effective tool. In a group of 50 patients with temporal lobe epilepsy, the CT scan demonstrated structural lesions in 24% of the patients whereas MRI demonstrated lesions in approximately 32%. The same paper showed that the CT scan demonstrated unilateral temporal lobe atrophy in 2/25 (5.7%) of patients compared with 15/31 (39.5%) examined with MRI. MRI is particularly effective in demonstrating the various lesions of cortical development, in particular cortical dysplasias. The CT scan imaging was relatively crude, especially that obtained from the early machines, as illustrated in [Figure 3]. MRI using epilepsy surgery protocols is now preferred to the CT imaging as the gold standard of imaging in the presurgical investigation of epilepsy.
However, MRI had other advantages; it can demonstrate the postoperative anatomy, giving useful feedback to the surgical team, and in some cases, it is helpful in demonstrating a lesion in patients who have failed previous surgeries. In addition, it can show gyral patterns and variations therein as well as estimate the volume of various parts of the cerebral cortex. By utilizing blood oxygen level dependent (BOLD) imaging, it is possible to show the anatomical location of functional areas of the brain such as the primary motor cortex and demonstrate the relationship between these areas and pathological lesions. It is also possible to record an EEG within the MRI scanner, although this is primarily of research interest. Locations within the cranium, previously found either by direct observation or by frame-based stereotaxy using radiographic data from ventriculograms visualized by air or contrast studies, can now be accurately approached either with open surgery or standard stereotactic techniques, utilizing radiographic data from MRI scans. Hence, where electrodes are being placed within or over the brain structures either for recording or stimulation, it is possible to place them accurately in conjunction with framed or frameless stereotaxy and to confirm their position during recording.
In terms of resective surgery, the ability to localize a structural lesion, to demonstrate the epileptic focus by neurophysiological assessments, to confirm the epileptic zone by functional imaging such as PET and SPECT, and by testing cognitive ability with appropriate neuropsychological tests, the surgeon is now able to precisely define the area of resection and to carry out this resection accurately with the assistance of MRI guidance. [Figure 4] contrasts the appearance of unilateral MTS on an air-contrast tomogram, MRI, and in a pathological specimen. [Figure 5] shows the MRI scans of a patient managed in the pre-MRI period who had two temporal lobe resections without benefit because the posterior temporal lesion was missed. Illustrative of the evolution of these processes is a paper from Paris in 2000. The paper describes the investigation of 28 patients with cortical dysplasia (CD) encountered between 1964 and 1995, using stereoencephalography (SEEG) and latter utilizing the MRI. They were able to identify the epileptogenic zone using the neurophysiological signature of CD and to correlate this with the pathology in the resected specimen and the size and position of the resected tissue. They found that, when there was complete resection of the epileptogenic zone or complete resection of the CD, the patients became seizure free. This was achieved in 64% of the patients, a remarkable achievement because most of these patients were operated on without the benefit of direct brain imaging, which was only available for 7 of these patients.
Functional methods in epilepsy surgery have a long history but without the support of modern technology, for example, stereotactically-guided brain lesions,,,, or temporal lobe disconnections described by Turner. However, efforts to operate on the brain to modify its behaviour rather than to destroy tissue became a more efficient proposition after the development of direct brain imaging and particularly MRI. In addition to lesioning, there are two other techniques which can be described broadly as disconnection and stimulation.
At this point it is worth mentioning that, in the 1970s and 1980s, newer methods of destroying pathological tissue were developed to avoid the complications of open craniotomy. In particular, radiofrequency lesioning, the position of which could be determined with brain imaging and where the size could be regulated by the temperature of the probe, size of the probe, and duration of passing current, made this methodology more acceptable; this has, since then, been used successfully to eradicate the existing pathology in inaccessible areas in a variety of conditions in both adults and children., In particular, it has been used in periventricular heterotopia  and hypothalamic hamartoma.,
Stereotactic radiotherapy was an alternative first proposed by Barcia-Solario  to destroy tissue, thereby avoiding the trauma and complications of open surgery. It was used to destroy the hippocampal tissue as an alternative to selective amygdalo-hippocampectomy. The latter was an operation which was devised by Wieser and Yasargil to try and avoid the problems related to memory, which sometimes followed the classical temporal lobectomy. The stereotactic procedure was described by Regis in 1999; however, the problems with late cerebral edema and the long wait of 2 years for the definitive result to be obtained were the disadvantages of this approach. The technique has also been used to treat a hypothalamic hamartoma.
The earliest serious disconnection procedure was a callosotomy, first proposed by Van Wagenen and Herren, on the basis of two oberservations: That EEG could be changed by section of the corpus callosum in animals; and, that when malignant tumours in the frontal lobe spread across the corpus callosum, epilepsy was better controlled. At first, the procedure was fraught with technical difficulties and a high complication rate. However, the work by the group at Dartmouth, New Hampshire, using the operating microscope, demonstrated that if the procedure was done in two stages, the complications were significantly reduced. Despite this, reports of neuropsychological complications were initially discouraging, especially when total section was attempted in one procedure; now these procedures are mainly of academic interest. The operation is now used sparingly with good effects, and a few groups use radiosurgery or serial radiofrequency lesions to effect the disconnection. The operation of anatomical hemispherectomy, first described by Krynauw, being a major procedure and having late complications, has been modified over the years by other procedures including the Montreal subtotal hemispherectomy and Adam's modified closure, which improved the situation; however, there was still a problem in operating on children where the most common indication was hemimegalencephaly requiring an early operation. This problem has been solved by transforming the surgery into a largely disconnective procedure. This was pioneered by Delalande in Paris and Schramm in Bonn., Finally, a new procedure called multiple subpial transection, devised by Morrell, relied upon the anatomical arrangements in the cerebral cortex whereby the functional fibres, for example in the primary motor cortex, travel radially to be incorporated in the internal capsule, and the fibres that transmit the epileptic discharges travel horizontally. The surgical technique, which consists of multiple small vertical cuts in the cerebral gyri, starting subpially and extending to the white matter border, respects this anatomy. This was proposed for use in the eloquent cortex and had wide application, although very few were able to replicate the initial results. It has been proven to be useful in the treatment of Landau–Kleffner syndrome, for which there are no surgical alternatives.
The increasing use of stimulation in epilepsy arises from two factors. They are the improvements in the size and reliability, and in the decreasing cost of the stimulating devices, along with experience with cardiac pacemakers; and the challenge of providing a different method of treatment for patients who are unsuitable for resective surgery. In addition, the required surgery was less invasive than other functional procedures, the electrodes were small and could be located accurately, and the stimulus was reversible as it could be turned off. The first practical cardiac pacemakers were manufactured and used in 1962–1963. However, it was many years before similar devices were used in the treatment of Parkinson's disease by Benabid and Pollack in 1998. Similar devices were then developed for use in epilepsy. There are two methods of stimulation, namely, the open loop in which the stimulus parameters are independent and uninfluenced by seizures or the neurophysiological data; and, the closed loop in which either the seizures or the phenomena connected with the seizures are used to trigger or modify the stimulation parameters.
A very early target was the cerebellar cortex. This was suggested by the work of Cooper and Snider which stated that the stimulation of the cerebellar cortex in animals could modify the EEG. Irving Cooper pioneered this operation; however, the results were unimpressive, and in the two controlled trials available, there was no effect. This procedure has now been abandoned.,
The most common form of stimulation in current use is stimulation of the vagus nerve in the neck (VNS), although recently, the auricular nerve has been used for transcutaneous stimulation. The rationale for VNS is complex and includes the wide distribution of vagus nerve afferents to many parts of the brain. The stimulation of these afferents can produce evoked potentials in a large number of brain regions due to which VNS can produce changes in the EEG. Thus, in an animal model using strychnine, the spike frequency can be increased or decreased by VNS depending on the parameters of the stimulating current. Woodbury showed in various animal models in the early 90s that VNS might be used as a clinical method. Early reports of the treatment were encouraging, and subsequent controlled trials EOS3 and EOS5 showed a significant effect with this method.,, Although this treatment is basically an 'open loop' one, there has always been the facility to produce an extra stimulus should the patients or their carer perceive the warning or the beginning of a seizure, thus introducing an element of 'closed loop.' In addition, there is now a new version which detects increases in heart rate associated with seizures and produces an extra stimulus; this has been shown to be effective in a significant number of patients who had a 20% or more increase in heart rate. Recently, transcutaneous stimulation of the trigeminal nerve via the ophthalmic division has been reported.
A number of intracranial targets have been used including the deep brain nuclei, hippocampal structures, and cortical areas stimulated from the surface. In the case of the deep brain nuclei, all the methods are 'open loop.' These nuclei include the caudate nucleus and the posterior hypothalamus (neither of them have been subjected to controlled trial in humans), and the thalamic nuclei (the stimulation of which have proved to be more fruitful). On the basis of the extensive connections of the anterior nucleus of the thalamus to the frontal and temporal lobes, experimental animal work, and encouraging results from previous small scale trials, Fisher set up a multicentre trial in the USA, known as the Sante trial involving 110 patients. The trial was a randomized double blind trial. The electrodes were implanted bilaterally, using framed or frameless stereotaxy; there was a significant reduction in seizures and a follow-up study at 5 years after implantation showed that the effect was sustained., The subthalamic nucleus, known to be involved in the nigral control of epilepsy, first described in 1980, has been considered. It is a well-recognized target for Parkinson's disease. However, although there are good theoretical reasons for using this target, there has only been one uncontrolled trial in humans. The centromedian nucleus of the thalamus has been reported extensively by Marcus and Fransisco Velasco. Their first report in 1987 suggested good results in 5 patients who were stimulated for 2 hours every day; the target was located using ventriculography and the Schaltenbrand atlas. By 2001, and in their last report in 2006, they were using evoked potentials in the frontal region to find the best position for the electrodes., A trial of this target was conducted by Fisher, which concluded that the stimulation did not make a significant difference and that more work was needed. Recently, Valentin et al., using the target, defined and optimized by stereotactic placement and evoked frontal potentials, have reported the results in patients with frontal lobe and generalized epilepsy. In the former group, there were mixed results; however, in the patients with generalized epilepsy, there was significant and uniform improvement.
There is one practical 'closed loop' system, the responsive neurostimulator system (RNS) system (Neuropace) in which four-lead electrode systems are implanted in one or two previously identified epileptic foci. These are connected to an implantable device, which is usually implanted subcutaneously in a depression in the skull. This device can be programmed to recognize the neurophysiological signature in the electrocorticographic recording from the focus. There is a report of a good result in a patient treated with motor cortex foci which illustrates the use of the system. Good results from a controlled trial using this device have been reported, although there has been some criticism.,, Research into the anatomy and characteristics of neural networks, both in the normal brain, and with the pathophysiology of the epileptic brain, will improve the efficacy and expand the scope of these techniques.
This review has traced the evolution of the pillars of resective surgery, namely, clinical semiology, neurophysiology both interictal and ictal recordings, and the verification of structural pathology by MRI. The use of these factors to offer and plan resective surgery reached its peak by the end of the 20th century. This review has not detailed the technical advances in surgery to which the operating microscope and frameless stereotaxy have made a serious contribution. [Figure 6] shows the neurophysiological rig for recording intraoperative electrocorticography in the 1970s. Note that paper was used as the recording medium which limited the analysis. [Figure 7] compares the operating theatre at the Maudsley Neurosurgical Unit in 1950 and a modern theatre, wherein the additional equipment is clearly visible. Functional surgery for epilepsy, practiced since the beginning has become well established in the 21st century, clearly facilitated by technical advances in stimulators. Methods of practice have changed with multidisciplinary teams dealing efficiently and effectively with the increasing number of patients in a changing landscape where the proportion of patients presenting for resective surgery is falling and those for other forms of treatment is increasing.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7]