J Neurosurg 60:269-275, 1984 Perioperative use of somatosensory evoked responses in aneurysm surgery LINDSAY SYMON, T.D., F.R.C.S., ALEXANDER D. WANG, M.D., 1ZABEL E. COSTA E SILVA, M.D., AND FaED GENTILI, M.D. Gough Cooper Department of Neurological Surgery, Institute of Neurology, The National Hospital, Queen Square, London, England v" Somatosensory evoked potentials have been recorded during 34 operations for intracranial aneurysm. The central conduction time (CCT), the time between the N~4 peak (recorded at C-2) and the N2o peak (recorded at the cortex) in response to median nerve stimulation, has been found to be increased by administration of halothane, by brain retraction, and by temporary vascular occlusion in some instances. Increase of CCT to more than l0 msec, or disappearance of the response was associated in two cases with postoperative neurological deficit, neither permanent. In a further case, prolongation of CCT was used as a guide to the repositioning of an imperfectly placed clip on a middle cerebral artery aneurysm. The value of the technique as a perioperative monitoring system is discussed. KEY WORDS cerebral ischemia 9 somatosensory evoked potentials 9 central conduction time 9 intracranial aneurysm halothane 9 brain retraction ~ T HE demonstration of a critical cerebral blood flow threshold for the maintenance of electrical activity in the primate cortex 2-4 indicates the potential of continuous electrical recording as a means of monitoring the adequacy of the cerebral circulation during surgery. The use of the somatosensory evoked potentials (SEP's) in the study ofaneurysm patients has been reported, 8 and this technique is now used in the operating room for repeated monitoring of cortical electrical activity during aneurysm surgery. Clinical Material and Methods Somatosensory evoked responses were recorded during intracranial surgery in 33 patients with 38 aneurysms that were clipped in a total of 34 operations. The patients included 13 males and 20 females, aged from 9 to 63 years. Twenty-six of the operations were carried out 2 days to 3 weeks following a subarachnoid hemorrhage (SAH), and in two other cases the SAH took place several months preoperatively. In six cases the operations were carried out for the obliteration of giant aneurysms. Three were compressing the visual pathways, and one was associated with an arteriovenous malformation. Two giant aneurysms were present in a patient who was operated on initially for facial pain from a giant basilar aneurysm and subsequently for a silent large middle cerebral artery aneurysm. J. Neurosurg. / Volume 60 / February, 1984 Somatosensory evoked responses were generated by stimulation of the median nerve at the wrist by square waves of 0.15-msec duration, and of an intensity determined at preoperative recording to be three times the subjective threshold and adequate to sustain a small thumb twitch. Bipolar stimulating electrodes were placed with the cathode 3 cm proximal to the anode, and the rate of stimulation was 3/sec. The technique was modified from that of H u m e and C a n t : and has been described previously. 8'1~ Recording electrodes were placed over the surface of the spine at C-2 and over the areas of the somatosensory cortex bilaterally, using the C3/4 position in the international 10/20 system. The frontal reference, Fpz, was used. Either 128 or 256 sweeps were averaged on a Digitimer D200 averager.* Central conduction time (CCT), defined as the time interval between the arrival of the impulse at the C-2 level and at the somatosensory cortex, representing conduction within the central nervous system itself, was used as the principal measurement, and the amplitude of the SEP's was also measured throughout the procedure. The averager, amplifier, stimulus generator, and the * Digitimer D200 averager manufactured by Digitimer, Welwyn, Hertfordshire, England. 269 L. S y m o n , A. D . W a n g , I. E. C o s t a e S i l v a a n d F. G e n t i l i oscilloscope were mounted on a mobile trolley and placed at the foot of the operating table. A group of preamplifiers were placed beneath the operating table, as close as possible to the patient's head, and the electrode wires were braided to avoid movement and to reduce the saturation of recording amplifiers from electrical and/or magnetic fields such as are generated by surgical instruments, the suction system, diathermy, headlight, television video recorders, or fluorescent lights. The patients had initial recordings performed on the day before operation and, on the day of operation, electrodes were fixed in place after the induction of anesthesia and shaving of the scalp. The position of these recording electrodes is shown in Fig. 1. The recording was carried out continuously while the patient was placed on the operating table, throughout the surgical procedure, and again in the immediate postoperative period. A further recording was made on return to the ward after the patient had regained consciousness. Preoperative clinical grading according to the Hunt and Hess scale 6 was carried out for each patient by one of us (L.S.). Ten patients were in grade 0 or I, eight in Grade II, five in Grade III, and 10 in Grade IV. There were 1 1 middle cerebral artery aneurysms, eight aneu- F1G. 1. Schematic diagram showing the reference electrode at the nasion (Fpz) and the recording electrode at the scalp (C3/4 in the international 10/20 system), placed well away from the small frontotemporal skin flap used for the aneurysm surgery. The recording from the neck electrode is not shown in the figure. rysms of the anterior communicating artery, 16 aneurysms of the internal carotid or the carotid/posterior communicating artery, and five aneurysms of the basilar artery. Results Effect o f Surgery on Conduction The effect of surgery on conduction through the central nervous system is shown in Table 1. Preoperative recordings in the 33 patients revealed a mean CCT of 5.8 _+ 0.5 msec, with a range of 5.2 to 7.0 msec. There was no difference in the total group between these recordings and those of the normal control group (20 normal subjects with a mean CCT of 5.4 _+ 0.4 msec) or of a group of SAH patients without aneurysms. When comparing the preoperative recordings in various grades, however, mean results ranged from 5.7 _+ 0.4 msec in Grade I to 6.0 __ 0.5 msec in Grade IV patients. There was a significant difference only between Grade IV patients and control data (p < 0.01). After the induction of anesthesia with thiopental and intubation, conduction slowed to a mean of 6.0 ___0.6 msec, which was not significantly different from the immediate preoperative data. The standard form of maintenance of anesthesia in this clinic was by inhalational halothane in concentrations varying from 0.5% to 2%. This resulted in moderate hypotension and prolongation of CCT in all 33 cases. The average CCT after the maintenance of a stable level of anesthesia was 6.7 ___0.7 msec, significantly longer than the preoperative recording (p < 0.01) and the recording during induction of anesthesia (p < 0.05). Figure 2 shows prolongation of CCT in response to a changing concentration of halothane. There was also a significant reduction in blood pressure during halothane anesthesia, with mean levels of blood pressure falling from 87.8 _+ 15.8 to 70.1 _ 12.1 m m Hg after the attainment of a stable concentration of halothane (p < 0.01). Opening the dura had no significant effect on conduction, but brain retraction caused significant CCT prolongation in a number of instances. Although the groups as a whole showed no significant change, CCT tended to be rather longer (mean 7.0 _+ 0.9 msec, ranging from 5.4 to 9 msec) after application of selfretaining retraction. In four patients, conduction time in the retracted hemisphere exceeded that in the control TABLE 1 CCT in both cerebral hemispheres at different operative stages* Hemisphere Induction Increase Open Brain Clipping Pre-op of Anesth. Halothane D u r a RetractionDissection Before After Reverse End of Regain Anesth. Operation Consc. affected side 5.8_+0.5 6.0_+0.6 6.7___0.7 6.8+0.7 7.0-+0.9 6.9+0.8 6.9+0.7 7.4+1,2 7.2+0.9 6.8+0.7 6.1+0.4 no. of cases 30 34 33 34 34 38 38 34 34 31 unaffected side 5.9_+0.5 6.1_+0.6 6.5_+0.7 6.7_+0.8 6.9_+0.8 6.8_+0.9 6.8_+1.0 6.8___0.9 6.9_+1.0 6.8+0.8 6.1_+0.5 no. of cases 30 30 29 30 31 30 27 30 29 31 * Values are means + standard deviations of central conduction times (CCT) in msec. Affectedhemisphere is the side of operation. 270 J. Neurosurg. / Volume 60 / February, 1984 Perioperative evoked responses during surgery hemisphere by more than 0.6 msec (Fig. 3). During the dissection of the aneurysm and its parent vessels, CCT again became gradually prolonged; the mean CCT during this stage was 7.1 _+ 0.8 msec, and ranged from 5.8 to 9.0 msec, although these values are not statistically different from earlier stages of the operation. The application of an aneurysm clip to 13 aneurysms in 12 of the 33 patients resulted in a significant prolongation of CCT, varying from 0.4 to 3 msec, with an average value of 7.4 +_ 1.2 msec. This was significantly different from the CCT recorded before clipping (p < 0.05). The remaining cases showed no change of CCT during this phase. Effects o f Temporary Clips Temporary clips were applied to major vessels in 15 patients: four to the terminal carotid artery, six to the middle cerebral artery, four to one or both of the proximal anterior cerebral arteries, and one to the internal carotid artery above and below the aneurysm neck and to the posterior communicating artery behind the aneurysm. In nine cases there was no detectable change in CCT, even when the terminal carotid, middle cerebral, or anterior cerebral arteries were clipped. In the other six cases, there was an immediate and significant prolongation of CCT by up to 3 msec within 3 minutes of clipping, and in one case the N2o peak could not be identified in the affected hemisphere after the application of three temporary clips, as described above (Fig. 4). Removal of temporary clips resulted in the recovery of conduction velocity to the immediately preceding level within 15 minutes, except in the single case of triple vascular occlusion when a recognizable potential appeared within 8 minutes, but the potential did not approach the preceding level until 40 minutes had elapsed (Fig. 4). A summary of several interesting cases is given below. Following the application of clips to the aneurysm, the standard surgical procedure was to elevate the blood pressure from a mean level of between 60 and 70 m m Hg to the patient's normal blood pressure, and to reverse the neuromuscular blockade used for respiratory paralysis. In general, this had no effect on CCT; however, in eight cases CCT became further prolonged during this phase of the procedure, in all cases but one after the development of spontaneous respiration. In the four cases where pCO2 values were recorded at this stage, each was above 40 m m Hg. Prolongation of between 0.4 and 1.2 msec lasted a short time only or until the patient regained consciousness. By the close of the procedure and the application of a head bandage, CCT had recovered to an average value FIG. 2. The effect of changing halothane (H) concentrations on central conduction time (CCT). The CCT increased from 6.6 to 8.2 msec when halothane increased from 1% to 4%, and returned to 7.2 msec when halothane was reduced to 0.75%. In the upper pair of traces, the scalp electrode is the upper trace. In the middle and lower pairs, the upper trace is from the neck electrode. FIG. 3. Effect of brain retraction: brain retraction prolonged central conduction time (CCT) in the operated hemisphere from 6.6 to 7.4 msec (left), with a 1.6-msec interhemispheric difference (right). Left: In each pair of traces the upper trace is the neck electrode. Right: Upper pair of traces is recorded from the affected hemisphere, the lower pair is from the unaffected hemisphere. In each pair, the upper trace is from the neck electrode. BP = blood pressure in mm Hg. J. Neurosurg. / Volume 60 / February, 1984 271 L. Symon, A. D. Wang, I. E. Costa e Silva and F. Gentili FIG. 4. Case 2. Angiograms showing a left carotid ophthalmic artery aneurysm and an irregular right posterior communicating artery aneurysm in both lateral (A) and oblique (B) views. A small right carotid ophthalmic artery aneurysm and a large aneurysm at the basilar bifurcation including the origin of both posterior cerebral arteries are shown in anteroposterior (C) and lateral (D) views. Through a right frontotemporal craniotomy the posterior communicating artery aneurysm was clipped. E: During surgery, central conduction time (CCT) became fiat for 8 minutes after a triple occlusion lasting 1 minute 20 seconds. The upper neck trace (which was constant throughout) and serial tracings from the scalp before the triple occlusion, just before release of the occlusion and 8, 15, and 18 minutes following release, are shown. BP = blood pressure in mm Hg. o f 6.8 _+ 0.7 msec, still significantly longer t h a n at i n d u c t i o n o f anesthesia (p < 0.01), but, by the t i m e the patients had regained consciousness in the recovery r o o m , the C C T h a d recovered to 6.1 _+ 0.4 msec, significantly shorter t h a n at the close o f the o p e r a t i o n (p < 0.01) a n d not significantly different from the preoperative levels. Amplitude M e a s u r e m e n t s M e a s u r e m e n t s o f SEP's during i n d u c t i o n o f anesthesia showed generally slight decreases in a m p l i t u d e which were less striking t h a n delays in c o n d u c t i o n (Table 2). Alterations in c o n d u c t i o n velocity could be detected at various stages t h r o u g h o u t the p r o c e d u r e before a measurable decline in SEP a m p l i t u d e was noted; however, during occlusion o f m a j o r vessels, as the c o n d u c t i o n t i m e b e c a m e prolonged, SEP a m p l i t u d e declined rapidly, a n d in some instances the potential could scarcely be detected. The ratio between the N20 a n d N14 a m p l i tudes disclosed no extra i n f o r m a t i o n except during occlusion, a n d the same was true o f c o m p a r i s o n between the two hemispheres. O n l y in the circumstances o f individual hemisphere retraction or occlusion o f the vessel on one side d i d the a m p l i t u d e o f the two tracings differ. M e a s u r e m e n t s o f SEP a m p l i t u d e were, in addition, c o m p l i c a t e d by incidental noise to a m u c h greater extent t h a n the m e a s u r e m e n t o f c o n d u c t i o n velocity, although with i m p r o v e d techniques o f recording in the operating r o o m it m a y well be that a m p l i t u d e measurem e n t s will have increasing value. TABLE 2 Effect of surgery on somatosensory evoked potential amplitude ratio (N2o/Nt4)* After Clipping Reverse Anesth. End of Operation affected side 0.82 +_0.40 0.73 _+0.28 0.81 _+0.39 0.74 _+0.13 0.82 _+0.54 0.81 + 0.37 no. of cases 27 24 19 22 22 23 unaffected side 0.74 _+0.41 0.74 _+0.28 0.81 _+0.39 0.82 _+0.54 0.94 + 0.67 0.80 + 0.37 no. of cases 26 24 19 22 21 23 * Values are means _+standard deviations in msec. Affected hemisphere is the side of operation. 0.84 _+0.40 23 0.84 _+0.40 23 0.93 _+0.38 20 0.92 _+0.38 20 Hemisphere 272 Induction of Anesth. Increase Halothane Open Dura Brain Retraction Dissection J. Neurosurg. / Volume 60/February, 1984 Perioperative evoked responses during surgery FIG. 5. Case 1. Angiograms revealing a giant left internal carotid ophthalmic artery aneurysm (A) which was occluded with a Sugita clip (B). During surgery, temporary clips were applied for 10 minutes to the internal carotid artery both in the neck and intracranially distal to the aneurysm. The central conduction time increased to 10 msec after application of the clips. C: A neck trace above and serial scalp traces, before carotid occlusion, 1 and 8 minutes after carotid occlusion had begun, and 2 minutes after release of the clips. Illustrative Cases Case 1 This 59-year-old woman had an 18-month history of progressive deterioration of vision in the left eye, with left optic atrophy. Her visual acuity was reduced to 3/60, with a large central scotoma. Angiography revealed a giant aneurysm of the internal carotid-ophthalmic artery. The CCT rose to 10 msec on application of carotid intracranial and extracranial clips (Fig. 5), and took 30 minutes to return to the pre-clip level after removal of the temporary clips. The patient had a diminished level of consciousness without clear hemispheric signs for 4 weeks. The clinical picture was complicated by a leak through the aneurysm clip into the interhemispheric fissure, which may have contributed to the depressed level of consciousness. The patient went on to make a full recovery within 2 months without hemispheric deficit. Vision in the left eye improved. Case 2 This 50-year-old woman had a history of SAH without coma 2 months prior to admission. Angiography showed multiple aneurysms. The N2o wave disappeared following temporary triple vascular occlusion during obliteration of a difficult posterior communicating artery aneurysm and basilar tip aneurysm at the same operation. She had an unidentifiable N2o peak for about 8 minutes after 1 minute and 20 seconds of vascular occlusion. There was an appropriate postoperative hemispheric deficit of dysphasia and left-sided hemiparesis which cleared slowly but completely over 3 months. This patient's tracings are shown in Fig. 4. Case 3 This 58-year-old woman had a 10-year-history of J. Neurosurg. / Volume 60 / February, 1984 FIG. 6. Case 3. Central conduction time (CCT) was prolonged up to 10 msec during surgery when inadvertent occlusion of an opercular branch of the middle cerebral artery occurred, and was reduced promptly on repositioning the clip. There was no detectable neurological deficit and a smooth postoperative recovery. In each pair of traces, the upper is from the neck electrode. BP = blood pressure in mm Hg. hypertension under regular medical control. She suffered a sudden severe headache 2 days before admission, and SAH was confirmed by lumbar puncture. Angiography showed an aneurysm on each middle cerebral artery. Computerized tomography showed blood in the left Sylvian fissure. There was no focal neurological signs, but she had mild neck stiffness and headache (Grade It). The left middle cerebral artery aneurysm was clipped 7 days after SAH. In the course of surgery, a clip was applied to inadvertently include an opercular branch of the middle cerebral artery. There was prolongation of the CCT up to 10 msec, but it then returned to normal on more appropriate positioning of the clip (Fig. 6). There was no detectable neurological deficit after the operation and the patient had a smooth recovery. 273 L. Symon, A. D. Wang, I. E. Costa e Silva and F. Gentili Discussion Operations for intracranial aneurysms carry risk to the perfusion of the neural tissue supplied by the aneurysm-bearing artery. Soon after SAH, these dangers are compounded by altered reactivity in the basal vessels ~ and by the diminished or abolished autoregulatory capacity that has been shown to follow soon after SAH. 7 In giant aneurysms the difficulty of handling these masses, often partially filled with clot, has led some surgeons to perform preliminary intra-extracranial bypass procedures, and has frequently necessitated temporary occlusion of major arteries while the lesion is manipulated, evacuated, and satisfactorily clipped. The demonstration in recent years of threshold relationships between blood flow and neuronal function 2-4'~~ prompted us to extend our techniques of electrical monitoring into the operating theater in an attempt to increase the safety of aneurysm surgery. The position of the indifferent and recording electrodes used for SEP recording in our hands is sufficiently far away from the small frontotemporal flap routinely used in surgery of the anterior circle of Willis, and SEP's may therefore be recorded throughout surgery without disturbing the operating surgeon. The operating room is, however, a hostile environment for recording of electrical conduction. The surgical manipulation, anesthetic agents, changes in blood pressure, and pCO2 combine to diminish the ease of recording, and the electrical recordings themselves are much influenced by electromagnetic fields generated by diathermy apparatus, operating lights, microscopes, and television cameras. Satisfactory SEP recording is possible with care, however; where bilateral responses can be recorded, unilateral conduction change is likely to result from surgical manipulation and may therefore be informative. Bilateral conduction changes relate more to anesthetic and metabolic factors. We have selected the Nj4 to N2o peak interval for monitoring CCT in view of our experience in the use of their parameters in pre- and postoperative monitoring following SAH. 9 The N~4 latencies throughout the operation were remarkably constant. In this study, we found the preoperative recording differentiated only Grade IV patients in terms of CCT, their conduction time being significantly longer than normal control times. In perioperative recording, the response of all patients to anesthesia and brain retraction was not influenced by clinical grade. In the course of our study, CCT emerged as a sensitive index of the halothane anesthetic technique: in all our patients anesthetized with this agent, CCT became prolonged proportionately as the halothane concentration changed. These data will be reported more fully elsewhere. Within the limits of blood pressure manipulation during surgery, blood pressure changes unaccompanied by changes in halothane concentration were not associated with significant CCT change. During reversal of anesthesia as blood pressure was raised, there was no significant shortening of CCT in relation to such ele274 vation of the blood pressure. During reversal of respiration support in eight cases, we noted prolongation of conduction time, possibly related to transient elevations of arterial pCO2, although insufficient measurements were made to confirm this factor. We found that conduction time could indicate excessive brain retraction and, although in the majority of our cases insertion of self-retaining retractors was unassociated with changes in conduction time, in four cases conduction time was prolonged when the retractors were inserted. An interhemispheric difference of more than 0.6 msec was seen in these cases, suggesting excessive retraction. Interestingly, CCT showed some prolongation immediately after the application of a clip to an aneurysm neck in 12 of the 33 patients. The aneurysms involved were two terminal carotid artery aneurysms, two posterior communicating artery aneurysms, six middle cerebral artery aneurysms, and three basilar artery aneurysms. The effect on CCT of the application of temporary clips was demonstrated in one of four cases of clipping of the terminal carotid artery and in four of six proximal middle cerebral artery clips. None of the four anterior cerebral artery clips caused prolongation of CCT nor did the application of clips to any anterior communicating artery aneurysm. In this regard it is worth noting that, of eight cases with temporary bilateral anterior cerebral artery occlusion, seven suffered no postoperative deficits and one showed signficant neurological disturbance. In this patient a very difficult anterior dissection required bilateral temporary occlusion of the A~ segments for 3 minutes 15 seconds, followed by the development of akinetic mutism. It appears, therefore, that while assessment of SEP's in response to median nerve stimulation may be a reliable index of ischemia occurring in the middle cerebral artery territory, it may be inadequate to assess ischemia in the anterior cerebral artery distribution. We are investigating the possibility of using leg stimuli from the posterior tibial nerve, with midline recording in position Pz of the international 10/20 system, to cover situations of this type. Throughout the investigation, careful notes of amplitudes of the SEP were made, but these were found of rather less value than measurements of CCT latencies. The SEP amplitude is easily disturbed by changing the position of the electrodes, by altering the intensity of the stimulus, and by the varying thickness of the scalp. Although these factors are more troublesome in attempted serial recording than when electrodes are in a constant position during surgery, amplitude is also more susceptible to electrical interference than is the latency of responses. Probably the most reliable amplitude measurement method is a comparison of the N~4 to N2o amplitude ratio throughout surgery and, although this in general showed a decrease of the N20 amplitude in relation to Nl4 with increased delay of the response, there were circumstances when the N20 amplitude transiently increased in a way that is difficult to explain at the present time, and the relationships parJ. Neurosurg. / Volume 60 / February, 1984 Perioperative evoked responses during surgery ticularly during temporary occlusion appeared more reliable with CCT measurements. The group of cases as a whole showed no postoperative neurological deficit. There was a discrepancy between a normal electrical recording and the postoperative deficit in one anterior communicating aneurysm case already mentioned, and in a further two cases there was a significant association between postoperative problems and perioperative CCT abnormalities. The recording of SEP's is practicable during intracranial aneurysmal surgery. It in no way interferes with surgery and, although careful recording techniques are necessary because of the associated interference, the effects of halothane, retraction, and application of temporary clips for occlusion of a major vessel can be monitored by the technique. Our evidence so far suggests that prolongation of CCT up to 10 msec is not likely to be associated with postoperative deficit in terminal carotid or middle cerebral artery aneurysm cases, but a more detailed analysis of the safety period for complete loss of the evoked response without sequelae must clearly await further experience. Acknowledgments The authors wish to thank Mr. Bert Morton and his colleagues for technical assistance, and Mrs. Judit Ashton for typing the manuscript. References 1. Allcock JM, Drake CG: Ruptured intracranial aneurysms - - the role of arterial spasm. J Neurosurg 22: 21-29, 1965 J. Neurosurg. / Volume 60 / February, 1984 2. Branston NM, Strong AJ, Symon L: Extracellular potassium activity, evoked potential and tissue blood flow. J Nearol Sei 32:305-321, 1977 3. Branston NM, Symon L: Cortical SP, blood flow and potassium change in experimental ischaemia, in Barber C (ed): Evoked Potentials. Lancaster, England: MTP, 1980, pp 527-530 4. Branston NM, Symon L, Crockard HA, et al: Relationship between the cortical evoked potential and local cortical blood flow following acute middle cerebral artery occlusion in the baboon. Exp Nenrol 45:195-208, 1974 5. Hume AL, Cant BR: Conduction time in central somatosensory pathways in man. Electroencephalogr Ciin Neurophysiol 45:361-375, 1978 6. Hunt WE, Hess RM: Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg 28:14-20, 1968 7. Symon L: Disordered cerebro-vascular physiology in aneurysmal subarchnoid haemorrhage. Acta Neurochir 41:7-22, 1978 8. Symon L, Hargadine J, Zawirski M, et al: Central conduction time as an index of ischaemia in subarachnoid haemorrhage. J Neurol Sci 44:95-103, 1979 9. Wang AD, Cone J, Symon L, et al: Somatosensory evoked potential monitoring during the management ofaneurysmal SAH. J Neurosarg 60:264-268, 1984 10. Wang AD, Symon L, Gentili F: Conduction of sensory action potentials across the posterior fossa in infratentorial space-occupying lesions in man. J Neurol Neurosurg Psychiatry 45:440-445, 1982 Manuscript received June 16, 1983. Address reprint requests to: Lindsay Symon, T.D., F.R.C.S., Gough Cooper Department of Neurological Surgery, Institute of Neurology, The National Hospital, Queen Square, London WC1N 3BG, England. 275