Acta Neurol. Scandinav. 53, 182-188, 1976
Department of Neurology, University Hospital, Lund, Sweden.
THE ELECTROCARDIOGRAM IN YOUNG ADULTS WITH ISCHAEMIC STROKE BENGTHINDFELT and OLLENILSSOX ABSTRACT Electrocardiograms in young adults w i t h ischaemic stroke were analysed for abnormalities within 24 h after t h e onset of symptoms. Fifteen patients out of 44 had some ECG abnormality, mainly ST-T changes, sinus tachycardia and minor disturbances of intraventricnlar conductance. I n most cases t h e ECG abnormality could b e attributed to some extraccrebral cause. I t is concluded t h a t ischaemic brain lesion p e r se never, or very rarely, affects t h e electrocardiogram.
Cerebrovascular disease is frequently associated with electrocardiographic changes ( W i e d l e r 1974), supposed to be secondary to the brain injury itself ( M a n n i n g &? Cotton 1962, G u n n et al. 1968). T h e electrocardiographic abnormalities m a y be of different kinds depending on the type of lesion, Electrocardiographic patterns suggestive of myocardial infarction have been repeatedly observed with subarachnoid haemorrhage ( C r o p p & Manning 1960, Greenhoot & Reichenbach 1969). Raised intracranial pressure is associated with U-waves, ST-T changes and notched T-waves ( J a c h u c k et al. 1975). Ischaemic stroke is known to carry a high incidence of arrhythmias ( R e i n s t e i n et al. 1972), T-wave changes and a left ventricular strain pattern ( F e n t z & Gorinsrn 1962, Lavy et al. 1968). Cerebrovascular disorders occur predominantly in elderly people and especially i n patients with pre-existing cardiac disease.Consequently, a clinical study of electrocardiographic abnormalities associated with strokes in a n unselected group of patients will be biased by pre-existing cardiac dysfunction and will not reveal the true incidence of neurogenic influences on the electrocardiogram. Because of this, we decided to analyse the electrocardiograms recorded i n young adults without known cardiovascular disease who were suffering from acute ischaemic stroke.
MATERIALS AND METHODS Forty-four patients, all below t h e age of 40 a n d w i t h ischaemic supra- o r infratentorial infarcts, were studied w i t h respect to ECG changes. T h e patients were admitted to the Department of Neurology, University Hospital of Lund during the period 1970-1974. The diagnosis was based upon clinical criteria, and i n every instance the diagnosis was confirmed b y selective angiography. A l u m b a r puncture was always done, and t h e cerebrospinal fluid was normal w i t h respect to cell count and colour i n all patients included i n t h e present series. Thus, patients with evidence of haemorrhagic stroke were excluded. The patients were examined a n d a n ECG was recorded within 24 h of the stroke. The ECG recording was t h a t used routinely, i.e. a 12-lead ECG consisting of standard and extremity leads (1-111, aVF, aVR and aVL) and t h e precordial leads V,-V,. Serum transaminases, GOT and GPT, were measured over t h e subsequent 3 d if the ECG showed any kind of abnormality, or if the history suggested any cardiac disorder. In order to find out whether there was a correlation between the occurrence of ECG changes and t h e neurological deficit (Figure 3 ) , a simplified scoring system was used ( H i n d f e l t 1975 b ) , taking into account only m a j o r disabilities (speech disturbances, defects i n mentation, hemi- o r monoparesis, sensory defects, visual field defects and incontinence). The defects were separated into slight, moderate and severe, and were given t h e scores 2, 4 and 6, respectively. Visual field defects and incontinence were scored 2. I n cases of monoparesis analogue scores of 1, 2 and 3 were used to denote t h e degree of paresis. Of the 44 patients studied 31 were male and 13 female. The mean age was 29.4 years, with ages ranging from 1 6 to 39 years.
RESULTS
Among the 44 patients only 15 had abnormal ECG tracing; the rest were normal. T h e various ECG changes observed a r e shown in Table 1 . The most common abnormality was ST-T changes, consisting of depressed ST segments and T-wave changes over the left ventricle. Eight patients had this type of abnormality, four of whom were known to be hypertensive (arterial blood pressure above 150/90 inmHg). Altogether six patients were hypertensive; four had ST-T changes, one had clinical evidence of acute myocardial infarction (with a significant rise in serum transaminases) and one patient fulfilled the electrocardiographic criteria of a Wolff-Parkinson-White syndrome. Among the patients with pathological ECG’s, one patient had mononucleosis, complicated by myocarditis. Another male patient had a n ischaemic stroke secondary to thrombotic purpura. His father and twin brother were both known to have suffered from myocardial infarction. This patient had a n ECG with depressed ST segments. Consequently, only eight patients out of 44 had some ECG abnormality without any known predisposing cause. All of these changes were clinically benign, consisting of ST-T changes, minor disturbances
184 Table 1. The various kinds o f electrocardiographic abnormalities encountered among young adults w i t h acute ischaemic stroke.
No. of patients Arrhythmias Sinus tachycardia Sinus arrhythmia Conduction disturbances Intraventricular ST-T wave changes Miscellaneous Myocardial infarction hlyocarditis Wolff-Parkinson-White syndrome
2 1
3
a
a@ sin
dx
$o’
O?’.
00
00
.-
\fi oooooooor ‘0000 o 000 000 ‘10 0 # v
0
Figure I . The number of normal and pathological ECG’s w i t h infarctions w i t h i n uarious vascular territories. Filled circles denote abnormal tracings; open symbols denote normal ECG’s.
of intraventricular conductance and arrhythmias (sinus tachycardia (two cases) and sinus arrhythmia (one case). In Figure 1 the ECG findings have been correlated with the site of the lesion, i.e. the vascular territory of the infarction. Occlusions of either internal carotid artery have been depicted within the territories of the middle cerebral arteries, respectively. Despite that, infarctions within the region of the middle cerebral artery were most common; the lesions being more frequent in the left cerebral hemisphere (24 infarctions as compared to 14 in the right hemisphere). Infarctions were seen within all major vascular territories except the left anterior
185
dx
sin
Figure 2. The various kinds o f ECG changes encountered w i t h lesions within the various vascular territories. Connotations as follows : filled circles = ST-T changes, filled triangles = intraventricular conduction disturbance, filled square = WPW syndrome, open square = sinus arrhnthmia, and open, angled square = sinus tachycardia. Combined symbols indicate that both abnormalities were present.
cerebral artery. The graph does not demonstrate any significant differences with regard to the relative frequency of observed EGG abnormalities within the various regions. In Figure 2 the type of ECG abnormality has been compared with the site of the lesion. Two patients, included in Table 1, have been omitted, i.e. the patient with myocardial infarction and the patient with myocarditis secondary to mononucleosis. In these two patients the cardiac disorder may well have been the main predisposing factor to the cerebral infarction. The graph indicates that two patients with left-sided supratentorial lesions had sinus tachycardia ( a pulse rate exceeding 100 beats/min) , but it can be questioned whether this kind of cardiac disturbance should be considered pathologic with regard to what has happened to the patient. One patient with a lesion within the posterior cerebral artery had sinus arrhythmia, a finding which is not rare among healthy young adults. Minor intraventricular disturbances of conduction and ST-T changes were seen with lesions within both cerebral hemispheres. The occurrence in one patient of a WPW-syndrome with brainstem infarction should be considered an unspecific finding. In a large series of patients with infratentorial infarctions this type of ECG abnormality was rarely observed (Hindfelt 1975 a ) .
186 Number
of patients
a , 6
, 0
4
.
0
0
2
0
0
0
0
0
0 0
e0.0 0
0
0
.
.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
8
10
12
14
16
18 20 22 24 Disability score
0
0
0
. 0
2
4
26
Figure 3. The correlation between ihe neurological disabilifg score and the occurrence of abnormal ECG tracings, Filled and open sgmbols denofe abnormal and normal ECG’s, respectivelg.
The correlation between the occurrence of ECG changes and the neurological deficit is shown i n Figure 3. Although the two patients with maximal disability scores had ECG changes (depressed ST segments in both cases), it is evident t h a t there was no close connection between the extent of the neurological deficit and ECG changes.
COMMENT
T h e present study allows some conclusions to be reached on t h e relationship between brain function and cardiac activity in cerebrovascular disease. F r o m everyday experience we know that the intact brain continuously influences cardiac activity, and consequently m a y change the electrocardiogram. There is ample clinical and experimental evidence t h a t a n increased intracranial pressure may drastically affect cardiac activity as well as the electrocardiographic appearance (Jachuck ef al. 1975). But it may be questioned if brain injury p e r se, unaccompanied by seizure activity o r raised intracranial pressure, does affect the electrocardiogram. Thus, Lauy ef al. (1968) have suggested that the ECG changes observed in a n unselected patient population with ischaemic stroke primarily reflect cardio-vascular disease that predispose to cerebral infarction. I n a recent study on ECG findings with infratentorial infarctions (Hindfelt 1975 a ) , it was concluded t h a t associated ECG changes were merely coincidental and did not reflect the ischaemie lesion within the central nervous system. The present study is based upon a selected group of patients in whom the frequency of cardio-vascular disorders should be minimized. This is reflected in the overall low rate of abnormal ECG tracings, 15 out of 44 patients ( 3 4 per c e n t ) . I n unselected groups of patients with ischaemic stroke about 70 per cent of the recordings are abnormal
187 (Lauy et al. 1968, Hindfelt 1975 a ) . Thus, there is a n increase in the frequency of ECG changes with age, indicating that a t least a part of the ECG abnormalities are not related to the brain i n j u r y p e r se. In this material the number of pathological ECG tracings, not accounted for by some extracerebral cause (hypertension, myocardial infarction etc.) could be reduced to 8 out of 44 tracings, i.e. about 20 per cent of the patients. T h e number should probably be lower still since it may be argued that sinus tachycardia and sinus arrhythmia (synchronous with respiration) should not be considered pathological in young adults, especially not under the prevailing conditions. It should also be noticed t h a t there was n o case of atrial fibrillation or flutter, otherwise frequently encountered in patients with ischaemic stroke. The clinical findings in the patients with ECG changes did not differ from those observed in the majority of cases with normal ECG recordings. T h e lesions were within t h e same vascular territories and the neurological symptoms a n d signs were similar. There was no correlation between the neurological deficit and the frequency of ECG abnormalities. Consequently, the results suggest that the ECG changes accompanying ischaemic strokes should not be attributed to the brain damage p e r se. T h e ECG abnormalities a r e probably unrelated findings, though they may reflect the same underlying vascular disease i n progress. REFERENCES Cropp, G. J. & G. W. Manning (1960) : Electrocardiographic changes simulating myocardial ischemia and infarction, associated with spontaneous intracranial hemorrhage. Circulation 22, 25-38. Fentz, V. & J. Gormsen (1962) : Electrocardiographic patterns in patients with cerebrovascular accidents. Circulation 25, 22-28. Greenhoot, J. 11. & D. D. Reichenhach ( 1 9 6 9 ) : Cardiac injury and subarachnoid hemorrhage.--4 clinical, pathological and physiological correlation. J. Ncurosurg. 30, 521-531. Gunn, C. G., G. Sevelius, bl. J. Puiggari & F. K. Myers (1968) : Vagal cardiomotor mechanisms in the hindbrain of the dog and the cat. Amer. Physiol. 216,
258-262. Hindfelt, B. (1975 a ) : The electrocardiogram in infratentorial infarcts. J. neurol. Sci. 26: 251-257. Hindfelt, B. (1975 b ) : On the significance of subfebrility and fever in ischacmic cerebral infarction. .4cta neurol. scand. In press. Jachuck, S. J., P. S. Ramani, F. Clark & R. M. Kalhag (1975) : Electrocardiographic abnormalities associated with raised intracranial pressure. Brit. med. J. I , 242-244. La\ y, S., S. Stern, Y . Herishianu &L A. Carmon (1968) : Electrocardiographic changes in ischaemic stroke. J. ncurol. Sci. 7, 409-415.
Manning, G. W. & M. Cotton (1962): Mechanism of cardiac arrhythmia induced by diencephalic stimulation. Amer. J. Physiol. 203, 1120-1124. Reinstein, L., J. G. Gracey & J. A. Mine (1972): Cardiac monitoring of the acute stroke patient. Arch. phys. Med. 53, 311-314. Wiedler, D. J. (1974) : Myocardial damage and cardial arrhythmia after intracranial hemorrhage. A critical review. Stroke 5, 759-764. Received July 27, 1975.
Bengt H i n d f e l t , M.D., Department of Neurology, University Hospital, S-22185 Lund, Sweden.
Department of Neurology, University Hospital, Lund, Sweden.
THE ELECTROCARDIOGRAM IN YOUNG ADULTS WITH ISCHAEMIC STROKE BENGTHINDFELT and OLLENILSSOX ABSTRACT Electrocardiograms in young adults w i t h ischaemic stroke were analysed for abnormalities within 24 h after t h e onset of symptoms. Fifteen patients out of 44 had some ECG abnormality, mainly ST-T changes, sinus tachycardia and minor disturbances of intraventricnlar conductance. I n most cases t h e ECG abnormality could b e attributed to some extraccrebral cause. I t is concluded t h a t ischaemic brain lesion p e r se never, or very rarely, affects t h e electrocardiogram.
Cerebrovascular disease is frequently associated with electrocardiographic changes ( W i e d l e r 1974), supposed to be secondary to the brain injury itself ( M a n n i n g &? Cotton 1962, G u n n et al. 1968). T h e electrocardiographic abnormalities m a y be of different kinds depending on the type of lesion, Electrocardiographic patterns suggestive of myocardial infarction have been repeatedly observed with subarachnoid haemorrhage ( C r o p p & Manning 1960, Greenhoot & Reichenbach 1969). Raised intracranial pressure is associated with U-waves, ST-T changes and notched T-waves ( J a c h u c k et al. 1975). Ischaemic stroke is known to carry a high incidence of arrhythmias ( R e i n s t e i n et al. 1972), T-wave changes and a left ventricular strain pattern ( F e n t z & Gorinsrn 1962, Lavy et al. 1968). Cerebrovascular disorders occur predominantly in elderly people and especially i n patients with pre-existing cardiac disease.Consequently, a clinical study of electrocardiographic abnormalities associated with strokes in a n unselected group of patients will be biased by pre-existing cardiac dysfunction and will not reveal the true incidence of neurogenic influences on the electrocardiogram. Because of this, we decided to analyse the electrocardiograms recorded i n young adults without known cardiovascular disease who were suffering from acute ischaemic stroke.
MATERIALS AND METHODS Forty-four patients, all below t h e age of 40 a n d w i t h ischaemic supra- o r infratentorial infarcts, were studied w i t h respect to ECG changes. T h e patients were admitted to the Department of Neurology, University Hospital of Lund during the period 1970-1974. The diagnosis was based upon clinical criteria, and i n every instance the diagnosis was confirmed b y selective angiography. A l u m b a r puncture was always done, and t h e cerebrospinal fluid was normal w i t h respect to cell count and colour i n all patients included i n t h e present series. Thus, patients with evidence of haemorrhagic stroke were excluded. The patients were examined a n d a n ECG was recorded within 24 h of the stroke. The ECG recording was t h a t used routinely, i.e. a 12-lead ECG consisting of standard and extremity leads (1-111, aVF, aVR and aVL) and t h e precordial leads V,-V,. Serum transaminases, GOT and GPT, were measured over t h e subsequent 3 d if the ECG showed any kind of abnormality, or if the history suggested any cardiac disorder. In order to find out whether there was a correlation between the occurrence of ECG changes and t h e neurological deficit (Figure 3 ) , a simplified scoring system was used ( H i n d f e l t 1975 b ) , taking into account only m a j o r disabilities (speech disturbances, defects i n mentation, hemi- o r monoparesis, sensory defects, visual field defects and incontinence). The defects were separated into slight, moderate and severe, and were given t h e scores 2, 4 and 6, respectively. Visual field defects and incontinence were scored 2. I n cases of monoparesis analogue scores of 1, 2 and 3 were used to denote t h e degree of paresis. Of the 44 patients studied 31 were male and 13 female. The mean age was 29.4 years, with ages ranging from 1 6 to 39 years.
RESULTS
Among the 44 patients only 15 had abnormal ECG tracing; the rest were normal. T h e various ECG changes observed a r e shown in Table 1 . The most common abnormality was ST-T changes, consisting of depressed ST segments and T-wave changes over the left ventricle. Eight patients had this type of abnormality, four of whom were known to be hypertensive (arterial blood pressure above 150/90 inmHg). Altogether six patients were hypertensive; four had ST-T changes, one had clinical evidence of acute myocardial infarction (with a significant rise in serum transaminases) and one patient fulfilled the electrocardiographic criteria of a Wolff-Parkinson-White syndrome. Among the patients with pathological ECG’s, one patient had mononucleosis, complicated by myocarditis. Another male patient had a n ischaemic stroke secondary to thrombotic purpura. His father and twin brother were both known to have suffered from myocardial infarction. This patient had a n ECG with depressed ST segments. Consequently, only eight patients out of 44 had some ECG abnormality without any known predisposing cause. All of these changes were clinically benign, consisting of ST-T changes, minor disturbances
184 Table 1. The various kinds o f electrocardiographic abnormalities encountered among young adults w i t h acute ischaemic stroke.
No. of patients Arrhythmias Sinus tachycardia Sinus arrhythmia Conduction disturbances Intraventricular ST-T wave changes Miscellaneous Myocardial infarction hlyocarditis Wolff-Parkinson-White syndrome
2 1
3
a
a@ sin
dx
$o’
O?’.
00
00
.-
\fi oooooooor ‘0000 o 000 000 ‘10 0 # v
0
Figure I . The number of normal and pathological ECG’s w i t h infarctions w i t h i n uarious vascular territories. Filled circles denote abnormal tracings; open symbols denote normal ECG’s.
of intraventricular conductance and arrhythmias (sinus tachycardia (two cases) and sinus arrhythmia (one case). In Figure 1 the ECG findings have been correlated with the site of the lesion, i.e. the vascular territory of the infarction. Occlusions of either internal carotid artery have been depicted within the territories of the middle cerebral arteries, respectively. Despite that, infarctions within the region of the middle cerebral artery were most common; the lesions being more frequent in the left cerebral hemisphere (24 infarctions as compared to 14 in the right hemisphere). Infarctions were seen within all major vascular territories except the left anterior
185
dx
sin
Figure 2. The various kinds o f ECG changes encountered w i t h lesions within the various vascular territories. Connotations as follows : filled circles = ST-T changes, filled triangles = intraventricular conduction disturbance, filled square = WPW syndrome, open square = sinus arrhnthmia, and open, angled square = sinus tachycardia. Combined symbols indicate that both abnormalities were present.
cerebral artery. The graph does not demonstrate any significant differences with regard to the relative frequency of observed EGG abnormalities within the various regions. In Figure 2 the type of ECG abnormality has been compared with the site of the lesion. Two patients, included in Table 1, have been omitted, i.e. the patient with myocardial infarction and the patient with myocarditis secondary to mononucleosis. In these two patients the cardiac disorder may well have been the main predisposing factor to the cerebral infarction. The graph indicates that two patients with left-sided supratentorial lesions had sinus tachycardia ( a pulse rate exceeding 100 beats/min) , but it can be questioned whether this kind of cardiac disturbance should be considered pathologic with regard to what has happened to the patient. One patient with a lesion within the posterior cerebral artery had sinus arrhythmia, a finding which is not rare among healthy young adults. Minor intraventricular disturbances of conduction and ST-T changes were seen with lesions within both cerebral hemispheres. The occurrence in one patient of a WPW-syndrome with brainstem infarction should be considered an unspecific finding. In a large series of patients with infratentorial infarctions this type of ECG abnormality was rarely observed (Hindfelt 1975 a ) .
186 Number
of patients
a , 6
, 0
4
.
0
0
2
0
0
0
0
0
0 0
e0.0 0
0
0
.
.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
8
10
12
14
16
18 20 22 24 Disability score
0
0
0
. 0
2
4
26
Figure 3. The correlation between ihe neurological disabilifg score and the occurrence of abnormal ECG tracings, Filled and open sgmbols denofe abnormal and normal ECG’s, respectivelg.
The correlation between the occurrence of ECG changes and the neurological deficit is shown i n Figure 3. Although the two patients with maximal disability scores had ECG changes (depressed ST segments in both cases), it is evident t h a t there was no close connection between the extent of the neurological deficit and ECG changes.
COMMENT
T h e present study allows some conclusions to be reached on t h e relationship between brain function and cardiac activity in cerebrovascular disease. F r o m everyday experience we know that the intact brain continuously influences cardiac activity, and consequently m a y change the electrocardiogram. There is ample clinical and experimental evidence t h a t a n increased intracranial pressure may drastically affect cardiac activity as well as the electrocardiographic appearance (Jachuck ef al. 1975). But it may be questioned if brain injury p e r se, unaccompanied by seizure activity o r raised intracranial pressure, does affect the electrocardiogram. Thus, Lauy ef al. (1968) have suggested that the ECG changes observed in a n unselected patient population with ischaemic stroke primarily reflect cardio-vascular disease that predispose to cerebral infarction. I n a recent study on ECG findings with infratentorial infarctions (Hindfelt 1975 a ) , it was concluded t h a t associated ECG changes were merely coincidental and did not reflect the ischaemie lesion within the central nervous system. The present study is based upon a selected group of patients in whom the frequency of cardio-vascular disorders should be minimized. This is reflected in the overall low rate of abnormal ECG tracings, 15 out of 44 patients ( 3 4 per c e n t ) . I n unselected groups of patients with ischaemic stroke about 70 per cent of the recordings are abnormal
187 (Lauy et al. 1968, Hindfelt 1975 a ) . Thus, there is a n increase in the frequency of ECG changes with age, indicating that a t least a part of the ECG abnormalities are not related to the brain i n j u r y p e r se. In this material the number of pathological ECG tracings, not accounted for by some extracerebral cause (hypertension, myocardial infarction etc.) could be reduced to 8 out of 44 tracings, i.e. about 20 per cent of the patients. T h e number should probably be lower still since it may be argued that sinus tachycardia and sinus arrhythmia (synchronous with respiration) should not be considered pathological in young adults, especially not under the prevailing conditions. It should also be noticed t h a t there was n o case of atrial fibrillation or flutter, otherwise frequently encountered in patients with ischaemic stroke. The clinical findings in the patients with ECG changes did not differ from those observed in the majority of cases with normal ECG recordings. T h e lesions were within t h e same vascular territories and the neurological symptoms a n d signs were similar. There was no correlation between the neurological deficit and the frequency of ECG abnormalities. Consequently, the results suggest that the ECG changes accompanying ischaemic strokes should not be attributed to the brain damage p e r se. T h e ECG abnormalities a r e probably unrelated findings, though they may reflect the same underlying vascular disease i n progress. REFERENCES Cropp, G. J. & G. W. Manning (1960) : Electrocardiographic changes simulating myocardial ischemia and infarction, associated with spontaneous intracranial hemorrhage. Circulation 22, 25-38. Fentz, V. & J. Gormsen (1962) : Electrocardiographic patterns in patients with cerebrovascular accidents. Circulation 25, 22-28. Greenhoot, J. 11. & D. D. Reichenhach ( 1 9 6 9 ) : Cardiac injury and subarachnoid hemorrhage.--4 clinical, pathological and physiological correlation. J. Ncurosurg. 30, 521-531. Gunn, C. G., G. Sevelius, bl. J. Puiggari & F. K. Myers (1968) : Vagal cardiomotor mechanisms in the hindbrain of the dog and the cat. Amer. Physiol. 216,
258-262. Hindfelt, B. (1975 a ) : The electrocardiogram in infratentorial infarcts. J. neurol. Sci. 26: 251-257. Hindfelt, B. (1975 b ) : On the significance of subfebrility and fever in ischacmic cerebral infarction. .4cta neurol. scand. In press. Jachuck, S. J., P. S. Ramani, F. Clark & R. M. Kalhag (1975) : Electrocardiographic abnormalities associated with raised intracranial pressure. Brit. med. J. I , 242-244. La\ y, S., S. Stern, Y . Herishianu &L A. Carmon (1968) : Electrocardiographic changes in ischaemic stroke. J. ncurol. Sci. 7, 409-415.
Manning, G. W. & M. Cotton (1962): Mechanism of cardiac arrhythmia induced by diencephalic stimulation. Amer. J. Physiol. 203, 1120-1124. Reinstein, L., J. G. Gracey & J. A. Mine (1972): Cardiac monitoring of the acute stroke patient. Arch. phys. Med. 53, 311-314. Wiedler, D. J. (1974) : Myocardial damage and cardial arrhythmia after intracranial hemorrhage. A critical review. Stroke 5, 759-764. Received July 27, 1975.
Bengt H i n d f e l t , M.D., Department of Neurology, University Hospital, S-22185 Lund, Sweden.
