Epilepsy Surgery

Introduction

I used to be an old fashioned general neurosurgeon with a wide general practice covering all aspects of neurosurgery. Nowadays epilepsy surgery provides the mainstay of my elective surgical practice. It is probably the most rewarding surgery I have ever done. It is a service that can only be provided through the medium of the multidisciplinary team and I am privileged to be part of one of the best teams in the business. It is a field that is constantly changing as new technologies widen the hoirizons of trhe possible. Thd requirements of operating on brain that not only looks normal but which may have normal function makes epilepsy surgery the most challenging of all the neurosurgical disciplines that I have been involved in. I hope this section of the web site conveys the excitement I feel in being part of this service in what is becoming the twilight of my career.

History of epilepsy surgery in Bristol

Services like this don't happen overnight. When I started as a consultant in Bristol there was no epilepsy surgery service. My predecessor Michael Torrens had worked with Dr Jonathan Bird, a consultant neuropsychiatrist to do a small number of cases and establish the need for a service in the south west. My training in Liverpool and Manchester had included the overall management of epilepsy including exposure to some conventional epilepsy surgery. The need for a local surgical service for epilepsy was clear and in the absence of anyone else in the department to take on this challenge I began to work with Dr Bird.

The problem with starting a new service in the modern health service is that there is no mechanism that encourages service development. The truth is that the system resists all attempts to do this. Initially all we could do was to treat patients on an individual case basis. Once we had established a clinical need, we were able to put together a business case for a formal service which we did in the mid 1990’s and in 1994 this was supported by the trust.

For the first time we had the infrastructure for a proper adult epilepsy surgery service. The neuro-psychiatrist, Dr Jonathan Bird, supported by Dr Nigel Walton, clinical neuro-psycholgist provided the clinical assessment. Dr Bird established the principle of dual training in neurophysiology. We had access to MRI scanning of sufficient quality to identify temporal lobe structural abnormality and I brought to the table my experience of image guided surgery. We were able, quite quickly to establish a service for temporal lobe epilepsy achieving results that were comparable with international series – 60% cure rates for temporal lobe epilepsy compared with 70% in international series (Palm desert 1990).

There were two main reasons why our results at that time were below average. Firstly we were a new service starting out. It takes time to establish a first quality service. Secondly we had a peculiar case mix, because in the early days all our patients came from a neuropsychiatry source. Things changed when the team was enhanced by neurologists with a specialist interest in epilepsy, first Dr Kasia Sieradzan and secondly Dr Samden Lhatoo. Dr Lhatoo continued the precedent set by Dr Bird in that he was dually qualified as a neurophysiologist. His friendly but ruthless determination, inherited from his famous relative, Sherpa Tensing helped to establish the epilepsy surgery service fully. Winning a national health and social care award for innovation in healthcare finally brought the service to the undeniable attention of the trust management.  Under his guidance the service was extended to patients with extra-temporal epilepsy with all the problems that that entailed specifically with the use of subdural grids.

The Current service

Dr Lhatoo left Bristol to take up a post of Epileptology Professor in Cleveland, continuing to climb ‘his everest’ and was replaced by Dr Howard Faulkner, continuing the evolution to our current service. Clinical evaluation of patients is now carried out by our neurologists - Dr Kasia Sieradzan, Dr Howard Faulkner and our new epileptology colleague Dr Madhu Ramnamoorti .  All are dual trained in neurophysiology, specifically in stereo EEG. – (more on this below). Neuropsychiatric input is provided by Dr Monica Mohan, who is involved in the epilepsy surgery assessment of learning disability patients but who has a much wider remit than that of the retired Dr Bird, in that she provides invaluable psychiatric input to the whole  epilepsy surgery service, something that was lacking before. Neuropsychology services are provided by Helen Thorburn, Margaret Newsom and their team. We also have a Neuroradiologist with a specific interest in epilepsy (Dr Marcus Likeman). Through him we now have access to a wider spectrum of imaging – High field (3Tesla) MRI, MR Spectroscopy and functional MRI, CT- PET scanning, MEG scanning, interictal SPECT and when necessary ictal SPECT as well. We have two videotelemetry units, one dedicated to epilepsy surgery, but this remains the rate limiting step in the epilepsy surgery assessment process.

The service has a regional and national referral base. We treat adults with both lesional and non lesional intractable epilepsy. My paediatric colleague, Michael Carter provides the surgical input to a supraregional paediatric epilepsy surgery service at Bristol Children's Hospital and provides an interface with the adult service. Patients are assessed by the epilepsy physicians and then investigated according to protocol. This includes specialized imaging, Video telemetry and neuropsychology assessment including where necessary a WADA test. Cases are discussed at a weekly epilepsy surgery meeting. Where concordance between the clinical findings, the semiology and the radiology can be identified patients are waitlisted for resective surgery. The mean waiting time for surgery after MDT is currently 14 weeks.

In October 2011 the service was enhanced by the addition of robotic Stereo EEG to our surgical armamentarium (Renishaw). The major problem with the management of extratemporal epilepsy as I discovered providing a service for Dr Lhatoo were the risks associated with sub dural grid insertion. Quite apart from being a two dimensional solution to a three dimensional problem, the risks of major complications were too high to allow the technique to be sustainable in the long term. Try as I might I was unable to reduce the risks of major complications, such as haematoma,infection and stroke to below 10%, an order of magnitude above other elective intracranial neurosurgery. Some aspects of this were outside my control – if a barrier nurse bed is used for general surgical emergencies then the risk of infection will be increased but others simply reflected the enormity of the surgery. Bilateral large craniotomies inevitably carry more risk than routine unilateral minimally invasive operations.

Robots have been around a long time in neurosurgery, as long as image guidance. However this has been technology looking for an application until now. Determined to find another solution to the use of subdural grids, I was very pleased to meet Professor Chico Cardinale, in Milan who has been using the surgical robot to place arrays of deep brain electrodes to record seizure onset in extratemporal epilepsy for a decade. By precise planning of electrode trajectories through the skin and skull he was able to position a three dimensional array of electrodes in the brain tailored to the hypothesis of epilepsy propagation put forward by his epileptologists. The procedure had minimal risk in his hands and had the additional advantage of being so minimally invasive that it did not interrupt the normal seizure frequency of the patients being recorded.

For the second time in my professional career, new technology changed my practice forever. Since the inception of Stereo EEG in my department we have stopped using sub dural grids altogether. We have now placed a total of 793 electrodes in 61 studies (average 13 per case - range 6-20) in 57 patients.   The procedure is still not without its risks - we have had two surgically significant haematomas but no infection (2.5%) - all in all, an order of magnitude less than with subdural grids. However we have had one epilepsy related death – SUDEP, following a drug reduction induced seizure (unrelated to the SEEG process itself)

Personal Activity

My input to the service has been to provide the sole surgical input to the epilepsy team. However the service is now generating a caseload that will support two adult surgeons with input from my paediatric surgery colleague, Mike Carter to help manage the ‘interface’ cases – teenagers promoted to the adult service. The establishment of the service coincided with the arrival of image guidance in the department. All our epilepsy surgery has therefore made use of this technology allowing for a minimally invasive approach to temporal lobe and lesional surgery, the facilitation of functionally guided surgery for extratemporal epilepsy and recently the use of surgical robotics

In the early days of the service I carried out a very minimally invasive procedure removing only part of the hippocampus, the inner part of the temporal lobe. Although this was an approach favoured by other epilepsy surgeons worldwide (Olivier – Montreal) I soon abandoned it as we had a number of patients with early recurrence of their epilepsy including two patients with apparent cure of their epilepsy that died suddenly of presumed SUDEP. I therefore reverted to a standard selective amygdalohippocampectomy for epilepsy of temporal lobe origin (Modified Spencer operation). I have never favoured the 'Yasargil' trans-sylvian approach myself as the aproach does not lend itself to image guidance, which, if used correclty, I find an invaluable adjunct to the Spencer procedure.

A disadvantage of the Spencer approach over the Yasargil is the higher incidence of visual field defect with this technique. This is rarely more than a partial upper quadrantinopia but in a third of my cases is enough to prevent someone regaining a driving licence which is often a goal of patients putting themselves forward for temporal lobe surgery. We are currently working with Professor Liam Gray in Cradiff comparing results of the two surgical approaches. I was fully prepared to change my surgical approach if the results favoured the Yasargil approach but the goal posts have been moved again by the devleopment of clinical useful tractography. This is sophisticated image processing that allows the reconstruction of white matter tracts in the brain. Because of the acute angles involved in the analysis of Meyer's loop, the intratemporal visual white matter tract this has proved difficult to image. Difficult that is until my senior registrar, Mr Angelo Pichhierri, found a way to produce clinically useful images allowing us to plan a trans-cortical approach to the medial temporal lobe that have allowed us to continue to use the Spencer approach without causing a visual field defect.

The introduction of SEEG has allowed two main groups of patients to be included in surgical treatment - patients with complex temporal lobe epilpesy and patients with extratemporal onset of their seizures. I am truely excited by being able to deliver precision, knowledge, based surgery first mooted by Jean Taileraich in the 1970’s when the technology did not exist to fulfil his dream of a multimodality stereotactic ‘morphogramme’ for each patient. It is a privilege to be able to live his dream! I recently presented the results of our SEEG cases in Japan. Here is a link to the data slides from that presentation - documents/seeg-rsults.pptx

Much of the extratemporal surgery is done under local anaesthesia. Below is a description of an awake craniotomy. Details of SEEG technique have been published and is available on line (click here)  

Vagus Nerve Stimulation

Where a focus for seizures cannot be identified we have offered patients non curative surgery in the form of Vagus nerve stimulation. This is a contentious technique not supported by many epilepsy surgery units in the UK . A further criticism has been made that we have used the technique too freely in place of more detailed investigation of extratemporal epilepsy. In response to the first criticism that VNS does not work I want to go on record as stating that no one was more skeptical of the technique when it was introduced than I was. I was effectively forced to offer the surgery to our worst affected patients by public opinion and the patient support groups. My cynicism relating to the hard sell techniques of the American company supplying the stimulators was overcome by the fact that my end stage patients improved! The improvement was not dramatic – there were no long term cures, but at least 50% of patients were more than 50% improved, which is the criterium required to introduce a new drug. A recent patient satisfaction survey asking patients to scale the effectiveness of their VNS on a scale of 0 - 10 produced the following results. 30% scored the treatment as 1 -2 (no good), 30% scored the treatment as 5 -6 (some use) while the rest scored the treatment as 8 - 9 (useful but nor curative). On this basis it was hard to deny patients with intractable epilepsy the chance to improve their epilepsy control. The use of VNS did not preclude someone from undergoing further evaluation in the future. Recent publication of our results (Ching et al 2013) has shown the benefit in terms of seizure control as being as good as any new drug with the benefit maintained with time.

However I retain the same cynicism regarding the introduction of the new cardiac gated VNS technology. Over the years the price of a VNS has been equivalent to the 5 year cost of a new drug. The original batteries (101) lasted between 5 and 15 years (mean 9.8 years) which compared favourably with drug costs at the time. New smaller batteries (103) were then introduced whose battery life was a lot shorter - good for share holders of Cyberonics but less good for patients and the cash strapped NHS! Having said that in our adult practice the mean battery life was still 4.9 years. Nevertheless to improve the cost effectiveness of the treatment we moved to the larger 105 batteries, in theory, returning the mean battery life to that of the original 101 series. In addition we halved the pulse width setting on the stimulator,  doubling the battery life again, without any demonstrable loss of benefit in our patients. It therefore came as no surprise to me to find that the company introduced new technology that had the potential to reduce battery life again - where the stmulator is activated automatically in response to tachycardia. The clinical evidence for the efficacy of this was borderline to say the least but the potential for reducing the battery life considerable. The price negotiated for the new 106 technology was on a par with the old 105 batteries so we moved over to the new technology. 

This situation has now evolved further. As of February 2018, there is still no independent data on the efficacy of the 106 system over the 105 (although we have some anecdotal indication that cardiac gating might be beneficial in some of our patients - direct communication with our epilepsy nurses). Nevertheless the price of the 106 batteries had gradually increased and now the 105 batteries have been withdrawn from the market in the UK altogether! The reason given was that our unit was the only one still using 105 batteries. No refund has been offered for remaining 105 units on our shelf despite the unilateral action of the company. In addition newer versions of the 106 system are being marketed, all with more complex programming possibilities (and with that the probabiltiy of reduced battery life!). I have refused to consider using these until we have established the place for the 106 batteries and we are able to produce reliable data on battery longevity independent of the company. Having said that I would consider evaluating the new technologies if the company bore the risk in terms of reduced battery life but there has been no indication that they are prepared to consider this. It is not their policy to invest in R&D in Europe. 

Epilepsy operations

Image guided selective medial temporal lobe resection

The commonest cranial operation for epilepsy is temporal lobe resection. Traditionally a full temporal lobectomy has always been done on the non dominant (usually the right) hemisphere with more selective techniques being reserved for the dominant hemisphere that controls speech and memory. The photo of the cadaveric brain demonstrates the difference. The blue highlighted area represents the extent of a conventional temporal resection. The red area is the extent of a selective operation. The small white arrow outlines the image guided approach to the medial temporal lobe. I have always used image guidance for epilepsy surgery and as a result I seldom carry out a full temporal lobectomy in non tumour cases. The technique I use I first described in 1992 when it was the subject of a ‘Tomorrow’s world’ program. The procedure is done under general anaesthetic. Once the patient is asleep, the image guidance is set up. (temporallobe1) We now use the Medtronic SNT StealthStation. This allows us to plan a minimally invasive approach across the temporal lobe.(temporallobe2) The medial structures of the temporal lobe, the Pes Hippocampus, the source of seizures of temporal lobe origin are identified (temporalobe3). The operation then consists of carrying out an en bloc resection of the mid hippocampus for histology followed by a resection of the anteromedial temporal lobe, the amygdala and the posterior hippocampal tail. The lateral temporal structures are preserved in this procedure, thereby preserving speech and memory function.

Awake craniotomy for low grade glioma.

There is no consensus as to the best management of a patient with a radiological diagnosis of a low grade tumour who is clinically well. Early surgery serves to establish a tissue diagnosis and remove the bulk of a tumour. However there is no evidence that this prolongs survival in the long term or is better management than the conservative approach of monitoring the progress of a tumour with regular scanning. The primary indication for surgery in this situation is often patient choice. There are as many patients who prefer to know what is going on and who wish to get on with their lives knowing that the bulk of their tumour has been removed as there are those who prefer to avoid surgery if at all possible. In the past clinicians took a paternalistic approach and decided for their patients. In broad terms a neurologist would decide against surgery whereas a surgeon would offer surgery. The modern approach is to recognize that the doctor is the servant of the patient and that our role is make sure that our patients are fully informed of the situation regarding their condition and to support them in the approach to their condition that they wish to take.

If it is decided to operate in this situation then every effort must be made to do so safely. For tumours arising in cortex that might control speech or movement the most sophisticated way of assessing a patient’s function during surgery is to monitor it directly i.e. do the operation under local anaesthetic. With modern anaesthetic techniques this is more straightforward than it sounds. This is the technique developed in our department with Dr John Carter, Consultant Anaesthetist.

The set up is more elaborate than for a procedure under general anaesthesia. First of all we do not give any premedication as we require the patient to be fully awake during the procedure. We take great care to position the patient carefully, usually on their side so that their airway is protected for the portion of the operation that they are asleep. The only pain sensitive part of the head is the skin so local anaesthetic is applied liberally to the places where the points of the head fixator penetrates the skin to hold the skull and to the incision site itself. Image guidance is always used so that a minimally invasive approach can be taken. We use ‘see through’ drapes so that we can watch the patient carefully throughout the procedure.

Once the patient is positioned, the head fixed, the image guidance set up and the incision site prepared, draped and anaesthetized, the patient is put to sleep using short acting anaesthetic agents. This is to allow the ‘carpentry’ to de done – the skull is opened with a high speed drill / saw system. This part of the procedure is not painful but it is a little noisy so I prefer patients not to be awake for this.

When the tumour has been exposed and I have orientated myself fully using the image guidance, the patient is the woken up, a process that takes about ten minutes. When they are fully awake I then proceed to resect the tumour, testing the patient’s speech and movement continually. When the resection is complete the patient is put back to sleep and the wound closed. If at any stage things become uncomfortable or the patient becomes concerned it is a very simple process to put them straight to sleep and proceed under general anaesthetic. 

Vagus nerve stimulation

This operation has the great advantage from the patient’s point of view that it does not involve any surgical intervention on the brain itself. The risks of permanent neurological deficit after surgery are therefore less. The operation is not curative but in our practice it produces significant improvement in some patients. The operation is carried out under general anaesthetic. A small 2 cm incision is made on the left side of the neck and the vagus nerve dissected out of the carotid sheath. The Cyberonics bipolar electrode is applied to the nerve which is the secured and tunneled to a subcutaneous pocket on the anterior chest wall where it is connected to the stimulator battery itself. The operation takes about an hour and the patient goes home the day after surgery. Two weeks later they are reviewed at Frenchay Hospital where the stimulator is turned on. It is programmed to produce a cyclical stimulation for 30 seconds every five minutes which the patient feels as a tightness in their throat. Over the first few weeks the strength of stimulation is slowly increased to therapeutic levels. The efficacy of the stimulation is then assessed over the next few months.

Audit results

An independent audit of the epilepsy surgery activity from 1990 to 2000 was carried out by Dr Sieradzan, Consultant neurologist when she joined the epilepsy surgery team. These results have been presented to the South west regional neurology audit meeting but have not been more widely disseminated. I reproduce some of them here. In addition I have added my own audit of epilepsy surgery from 2000 to 2003, from 2004 to 2013 and from 2014 to 2017. The latter data represents a further audit cycles in the evolution of the service. We now have input from three Consultant neurologists to the service which has changed the referral base of surgical patients, giving us a broader cross section of patients that are suitable for surgery. Secondly my surgical technique has changed in that I now perform a standard medial temporal lobe resection as opposed to the more minimally invasive operation that we performed originally. Thirdly we now have access to more sophisticated MRI scanning allowing a more accurate analysis of medial temporal anatomy. Finally SEEG has revolutionised the assessment of complex temporal and extratemporal epilepsy. I am aware that the data is not complete.

Activity

My interpretation of these results are as follows. It took us a while to achieve the international standard in terms of epilepsy control in temporal lobe epilepsy. Improving the referral base and changing the surgical technique wil have been two important factors. The introduction of SEEG has introduced a new element to this in that we are redefining our understanding of network epilepsy. I refer to one case where SEEG showed a 'perfect' localisation of seizure onset to the medial temporal lobe but resection made no difference to the seizure control. Repeat SEEG showed a clear network involving the insula - there is much to learn about epilepsy networks!

A comment about visual field defect after temporal lobe resection. My approach - a modified Spencer transcortical resection has produced a 30% rate of significant visual fiedl defect. By comparison the Yasargil, trans-sylvian approach is said to produce a far lower rate of defect, such that neither of my colleagues that use this technique routinely measure visual fields! Prof Gray and I are currently looking into this by comparing Bristol and Cardiff data. However we are now using tractography of Meyer's loop to plan the temporal lobe approach avoiding significant visual field defects so far.

SEEG has revolutionised our assessment of complex epilepsy but it has not been entirely without risk in our hands. We have had some case of haemorrhage assoicated with the placment of oblique eletrodes in patients with thick bone. The technique we introduced differed from that used in Milan in a number of ways that may have proved to be significant. Firstly we have not used 3D angiography for planning. Secondly we have not predrilled the bone to set the trajectory for the main drilling. Changing our technique to accomodate these differences is underway

Extratemporal lesional surgery presents the same challenges as oncological surgery as long as there is a clear demarcation between the pathology and the normal brain. However in cases where either there is no macroscopic differentiation between the target and normal brain or where the macroscopic changes are so subtle (dyplasias) that there are no anatomical cues to assist the surgery this surgery is technically very challenging. The margins of error are small particularly in areas of eloquent cortex such that conventional image guidance is unhelpful and may actually be misleading. Apart from cases where I want to montior speech during resection I no longer use awake craniotomy as I have found this technique to be misleading also. Patients may have no deficit during the procedure but develop one due to delayed ischaemia post operatively. Intraoperative neurophysiology is much more sophisticated than it used to be and more user friendly such that this can more helpful in monitoring motor function than by having the patient awake.

Increasingly we are making use of the accuracy of robotics to guide resection by placing guide cathetors along the same track of the SEEG electrodes and then tailoring the cortical resection to the cortex around the catheter. Intraoperatve MRI has not been a helpful adjunct in cases with no anatomical target.

The only truely accurate way to approach a target within the brain is not to open the head at all. This opens the door for minially invasive techniques like interstitial hypethermia, Convexion enhanced drug delivery, central neurostimulation and non surgical approaches like radiosurgery. These techniques will be the future of epielpsy surgery in my view

Deaths

The first thing to note about this data is the mortality risk of untreated severe epilepsy! The deaths that occurred due to epilepsy in patients who were being followed up after surgery were all classified as SUDEP – Sudden Unexplained Deaths in Epilepsy. The early patients were all fit free after surgery and had undergone minimally invasive selective temporal lobe resections. One had been fit free for a year, another for 3 years and another for 4 years before the SUDEP occurred. While this is likely to reflect the severity of the epilepsy suffered by our patients in the early days of the program it was hard for me to escape the conclusion that by restricting the extent of hippocampal resection as much as I did, I had succeeded in preventing regular seizures but not rare severe seizures. This was one reason why I reverted to a standard volume amygdalohippocampectomy. The last case was a sad case; a man with an occipital dysplasia who had multiple surgeries each with temporary success in reducing his seizure frequency. The final procedure was an SEEG guided topectomy using intra-operative MRI and robotic placement of catheters to localise the electrical focus. The aim was to preserve his visual fields. Initial improvement in his seizures was not maintained and he died a SUDEP death. Would a formal occipital lobectomy have been better? We concluded that it would not. Other SEEG recordings of occipital lobe epilepsy have revealed very active network - this is unlikley to have been focal epilepsy.

Vagus Nerve stimulation

We will have to wait and see how the battery life is with the 105 and 106 batteries. The two 105 batteries that we have replaced were in patients from the paediatric program where far higher parameters are used than we use in the adult service.  We continue to have a number of lead failures but removal for overall system failure is less than with the earlier systems. The thinner batteries are tolerated better than the original 101 battery. I will continue to resist changing the systems in the absence of concrete data showing soem benefit.

Final Comments

I have been privileged to be involved in the development of the epilepsy surgery service in Bristol since its inception in 1991. The service has been forced to evolve slowly but progressively throughout the 1990’s because of lack of proper funding. Despite this we have now reached the stage where we have an active multidisciplinary team in place with most, but not all of the resources to provide a full epilepsy surgery service. We still need dedicated neurophysiology consultant time, ring fenced in patient beds and functional imaging on site before we can say that we are as well resourced as other units. Our results are now comparable with the international standard. However we still need to continue to develop, in particular in the field of extratemporal epilepsy surgery. My aim is now to maintain and develop the epilepsy surgery service at Southmead.  Epilepsy surgery is one of the most rewarding things I do and it is a service I want to continue to provide in the South West. In the present political climate this may be easier said than done!