OUP user menu

Clinical patterns of neurological involvement in Behçet's disease: evaluation of 200 patients

Gülsen Akman-Demir, Piraye Serdaroglu, Banu Tasçi
DOI: http://dx.doi.org/10.1093/brain/122.11.2171 2171-2182 First published online: 1 November 1999

Summary

In order to define the patterns of neurological involvement in Behçet's disease and to assess prognostic factors, 558 files of the neuro-Behçet out-patient clinic were reviewed. Those patients without any evidence of objective neurological involvement as well as the patients with other possible explanations for the neurological picture, and cases not fulfilling the criteria for Behçet's disease were excluded. The remaining 200 cases (155 male, 45 female) were evaluated: 162 had parenchymal CNS involvement (brainstem or `brainstem +' involvement in 51%, spinal cord involvement in 14%, hemispheric involvement in 15% and isolated pyramidal signs in 19%) while 38 had secondary or non-parenchymal CNS involvement. In the first group the most common findings were pyramidal signs, hemiparesis, behavioural changes and sphincter disturbance, whereas in the second group the syndrome of raised intracranial pressure due to dural sinus thrombosis was the main clinical manifestation. In 60% of the cases with parenchymal involvement, CSF was hypercellular and/or had an elevated protein level, whereas in cases with non-parenchymal involvement the CSF was usually normal except for the elevated pressure. In more than half of the patients with parenchymal involvement, MRI showed brainstem and/or basal ganglion lesions. Forty-one per cent of the cases had a course with at least one attack and remission, another 28% also had attack(s) but showed secondary progression, 10% had primary progression and 21% had silent neurological involvement. Survival analysis was performed in patients who had at least a 3-year duration of neurological disease. Parenchymal involvement, elevated protein and/or pleocytosis in the CSF, `brainstem +' type involvement, primary or secondary progressive course and relapse during steroid tapering were all associated with a poorer prognosis.

  • Behçet's disease
  • neurological involvement
  • neuro-Behçet
  • CSF
  • prognosis
  • HLA = human leucocyte antigen
  • IgG = immunoglobulin G
  • IH = intracranial hypertension
  • MCA = middle cerebral artery
  • VA = vertebral artery
  • VCSS = vena cava superior syndrome

Introduction

Behçet's disease was first described by the Turkish dermatologist H. Behçet as a three-symptom complex comprising uveitis, oral aphtae and genital ulcerations (Behçet, 1937). Later, other features of the disease were established. Today we know that the disease is a multisystemic, recurrent, inflammatory disorder affecting the eyes, skin and mucosa, joints, vascular system (mainly the veins), lungs, gastrointestinal tract and nervous system (Inaba, 1989).

According to the latest diagnostic criteria for Behçet's disease (Table 1), recurrent oral aphtae are a prerequisite, accompanied by any two out of the following: genital ulcerations; skin lesions such as erythema nodosum, foliculitis and ulcerations; eye involvement such as anterior and posterior uveitis; and skin pathergy reaction, a peculiar skin hyper-reactivity to non-specific physical insults such as pinprick (International Study Group for Behçet's Disease, 1990).

View this table:
Table 1

Diagnostic criteria for Behçet disease (International Study Group for Behçet's Disease, 1990)

Recurrent oral aphtae: plus any two ofat least three times in a year
genital ulcerations:active lesion or scar
skin lesions:erythema nodosum, foliculitis, other ulcerations
eye involvement:anterior or posterior uveitis, or retinal vasculitis
positive pathergy test:skin hyper-reactivity to pinprick
(sterile pustule formed in 24–48 h)

Although not included in these diagnostic criteria, there are some other features commonly seen in patients with Behçet's disease which show geographical variation (O'Neill et al., 1993; Hamuryudan et al., 1998). These are thrombophlebitis, oligo-arthritis, gastrointestinal ulcerations and neurological involvement. Certain organs, such as the kidneys, heart and liver, seem to be relatively spared (Shimizu et al., 1979).

Vasculitis is usually considered to be the central pathological feature in Behçet's disease (O'Duffy, 1990; Ehrlich, 1997). However, a vasculitic process cannot usually be demonstrated in the CNS (Hadfeld et al., 1997). Studies on the pathology of the CNS involvement have shown that both a low-grade chronic lymphocytic or neutrophilic meningoencephalitis and multifocal necrotic foci, predominantly in the brainstem and basal ganglion region, are seen (Rubinstein et al., 1963; Sugihara et al., 1969). Although the tissues that are involved in Behçet's disease may show various types of histopathological lesions according to the age of the lesion at the time of examination, a non-specific inflammatory reaction with mononuclear and/or neutrophilic predominance seems to occur in common (Powell et al., 1991; Gül et al., 1995).

The aetiopathogenesis of Behçet's disease remains unknown. When the disease was first reported a viral aetiology was suggested (Behçet, 1937), and this was later investigated widely. Among the candidate viruses, herpesviruses received greatest interest (Eglin et al., 1982; Young et al., 1988). Streptococci have also been implicated (Behçet's Disease Research Committee of Japan, 1989; Mizushima, 1991). However, there is no evidence of a direct infectious cause. Antigenic cross-reactivity seems to be a better explanation, since antigens such as heat-shock proteins have been shown to be shared between micro-organisms and samples from patients with Behçet's disease (Lehner et al., 1991; Stanford et al., 1994; Tasçi et al., 1998), suggesting that Behçet's disease has an autoimmune nature (Sakane, 1997). However, there is opposition to this autoimmune theory based on facts such as its male predominance, the lack of concurrent autoimmune diseases, the lack of any specific antigen or antibody and the lack of any relationship with human leucocyte antigen (HLA) class II antigens (Yazici, 1997). Genetic susceptibility to Behçet's disease has been noted in certain populations. In the Mediterranean region and the Middle East, HLA B51 is significantly more common among Behçet's disease patients. For example, up to 84% of patients with Behçet's disease in Turkey are positive for HLA B51 (Yazici et al., 1980), a marker that is not found in other populations, such as British and North American patients (O'Duffy, 1994).

Neurological involvement in Behçet's disease was first described in 1941, and the first autopsy case was reported in 1944 (Berlin, 1944). Since then many series have been published. Here we present a clinical evaluation of our 200 patients with Behçet's disease and neurological involvement; to our knowledge this is the largest series ever published. Our aim is to define patterns of neurological involvement in Behçet's disease and to assess prognostic factors.

Patients and methods

Complete data from our neuro-Behçet out-patient clinic, which was first established in 1984, was re-evaluated retrospectively by four neurologists (G.A.D., B.T., Betül Baykan and P.S.). Earlier cases dating back to 1961 were added to this data. Re-evaluation of the total of 558 files for this out-patient clinic was carried out to determine whether individual patients had neurological involvement due to Behçet's disease. The files were further discussed in joint sessions. Patients without any evidence of objective neurological involvement, i.e. those with only headache or vague symptoms such as dizziness, who did not show any abnormality on neurological examination, were excluded, as were patients with other possible explanations for their neurological picture. Not fulfilling the criteria for Behçet's disease (International Study Group for Behçet's Disease, 1990) was also an exclusion criterion (probable neuro-Behçet cases). Among the 358 cases excluded, 37 had other causes underlying their neurological complaints, 23 cases did not have sufficient diagnostic work-up, 22 cases did not fulfil the diagnostic criteria for Behçet's disease, 186 cases had tension headache, migraine or psychoneurosis/depression with a normal neurological examination, and the remaining 90 cases had been sent for routine neurological examination without any neurological findings. Therefore, 200 files remained that were eligible for the final evaluation.

All cases seen after 1990 were sent for neuropsychological evaluation, and all cases seen after 1989 had at least one MRI scan.

In this evaluation, general epidemiological features, the frequency and distribution of other systemic findings, the characteristics of disease onset and course and the types of neurological involvement were noted. Prognostic factors were also evaluated in cases with a duration of neurological involvement of at least 3 years.

For statistical analysis, the χ2 test was used, correlation analysis was performed using Kendall tau-b, and the Kaplan–Meier test was used for survival analysis.

Definitions of terms used in the study

Attack = acute or subacute onset of new neurological symptoms and signs lasting >24 h which occurred at least 2 months after any previous attack. (Any worsening in the previous complaints during steroid tapering was not considered as a new attack if there were no new signs.)

Primary progressive course = slowly evolving and worsening neurological symptoms and signs over months or years with no preceding attacks.

Secondary progressive course = slowly evolving and worsening neurological symptoms and signs after at least one previous attack, or step-wise progression if there was more than one attack.

Silent neurological involvement = detection of abnormal findings on neurological examination in cases who did not have any neurological complaints except typical tension headache or migraine.

Disability status: 1 = self-sufficient, not dependent on another person or any aid; 2 = physically dependent (i.e. requiring help for walking or other motor activities); 3 = mentally dependent (i.e. severe, inappropriate behaviour or dementia requiring another person's care); 4 = both physically and mentally dependent; 5 = dead (due to any cause related to Behçet's disease).

Results

There were 200 cases who had neurological involvement; 155 were male and 45 were female (M : F = 3.4). Their mean age at admittance was 33.4 ± 9.1 years (median = 33). Age at onset of Behçet's disease was 25.8 ± 7.8 years (median = 25 years) and age at neurological onset was 31.5 ± 8.9 years (median = 31), with a mean interval of 5.6 ± 5.5 years between these two ages. In 15 cases (7.5%), neurological onset was concomitant with the onset of Behçet's disease; among the remaining six cases (3%), neurological onset preceded the onset of Behçet's disease by 1 year in three subjects and by 2, 5 and 9 years in the remaining subjects.

The whole group had a mean follow-up period of 41.8 ± 1.9 months. However, 44 cases had a single visit. The cases included in the survival analysis (103 cases with a duration of neurological involvement of >3 years and the seven patients who died within the first 3 years of neurological onset, a total of 110 cases) had a median follow-up of 59.8 months (64.7 ± 55 months) and the median duration of neurological involvement was 78 months (92 ± 57 months).

Systemic symptoms of Behçet's disease

All cases had oral aphtae, 188 cases had genital ulcerations (94%) and 166 had skin lesions (84%); eye involvement was present in 129 cases (66%). The skin pathergy test was positive in 83% of the patients tested. Other manifestations were as follows: arthritis or arthralgia, 56%; thrombophlebitis, 33%; arterial involvement, 3.5%; pulmonary involvement, 7%; gastrointestinal involvement, 3%.

Clinical patterns

According to the site of involvement in the nervous system, patients were divided into two main groups (Table 2), namely those with parenchymal CNS involvement and those with secondary or non-parenchymal CNS involvement. The latter group consisted of cases who had CNS dysfunction due to involvement of major vessels and those who did not show any CNS parenchymal involvement.

View this table:
Table 2

Sites of involvement of the nervous system

Parenchymal CNS involvement (n = 162)Secondary or non-parenchymal involvement (n = 38)
*These 31 cases showed only pyramidal findings with no other signs. MRI was available in 10 of these 31 cases. ECA = external carotid artery.
Brainstem involvement: 83 casesIntracranial hypertension: 34 cases
Isolated brainstem: 21 cases Dural sinus thrombosis: 20; VCSS: 1 case
`Brainstem +': 62 cases ?Aetiology despite MRI/angiography: 5 cases
Spinal cord involvement: 23 cases MRI/angiography not done: 8 cases
Isolated spinal cord: 10 casesAseptic meningitis + papilloedema: 1 case
`Spinal cord +': 13 casesArterial involvement: 3 cases
Hemispheric involvement: 25 cases VA dissection: 1 case; MCA occlusion: 1 case
Localization not possible: 31 cases* ECA aneurysm: 1 case

Parenchymal CNS involvement (n =162; 125 males, 37 females)

Onset.

Neurological onset was with an attack, which usually lasted a few days and showed a rather subacute evolution, in 102 cases (63%). In the remaining 60 cases (37%), slowly evolving signs and symptoms were evident. However, 14 of the latter group (23%) had at least one neurological attack later in the course. In all, 56 cases had one neurological attack, 29 had two attacks, 20 had three attacks and 11 had at least four attacks. In cases in which the duration of neurological involvement was >3 years, the median number of attacks was 1.5 in a median duration of 6.5 years.

Predominant syndrome.

Of the cases with parenchymal involvement, 83 had predominantly brainstem involvement (51%). However, only 21 of these had isolated brainstem involvement. The remaining 62 cases also had cognitive impairment and/or spinal cord disease (`brainstem +' type involvement). Twenty-three cases (14%) had predominantly spinal cord involvement (10 with isolated spinal cord involvement and 13 with additional brainstem and/or hemispheric involvement). Twenty-five cases (15%) had hemispheric involvement and in 31 cases the region affected could not be localized because there were only pyramidal signs (19%).

Clinical signs and symptoms (Table 3).

View this table:
Table 3

The most and least common neurological findings in the group with parenchymal CNS involvement

Most common (≥50% of cases)Less common (10–40% of cases)Least common (≥5% of cases)
Pyramidal signsBrainstem signsParoxismal episodes
(bilateral >2/3)Pyramidocerebellar syndromeSeizures
HemiparesisSensory disturbancePsychiatric disturbance
(>90% unilateral)FeverHearing loss
Behavioural changeParaparesisCerebellar syndrome
(1/3 apathy, 2/3 disinhibition)Meningeal signsOptic neuropathy
Sphincter/impotenceMovement disordersHemianopia
HeadacheExcessive daytime sleep/ hyperphagiaAphasia

In this group the most common finding was pyramidal signs (96%), which were bilateral in more than two-thirds of the cases. On the other hand, 60% had hemiparesis, 90% being unilateral. Behavioural changes, evolving over a few weeks or months, were notified by the families in 54% of the patients (one-third showed apathy and two-thirds showed disinhibition). Forty-eight per cent had sphincter disturbance and/or impotence. During an acute attack, 49% of the patients had headache.

Rather less prominent findings included brainstem signs (37%; 25 patients had various types of ophthalmoplegia other than internuclear ophthalmoplegia, three patients had internuclear ophthalmoplegia, six patients had bulbar signs, 10 patients had ophthalmoplegia and bulbar signs, 11 patients had other brainstem syndromes, and other combinations occurred in the remaining five cases); pyramidocerebellar syndrome (33%); sensory disturbance (29%; 16 patients had hemihypaesthesia, 15 patients only had impaired deep sensation, 13 cases had a transverse sensory level and three had patchy sensory loss—only three patients reported sensory complaints, the remaining were noted at the examination); and fever during an attack or episodes of high fever without any demonstrable infectious cause (19%).

Findings encountered less commonly were paraparesis (11%), meningeal signs (8%), movement disorders such as hemichorea, hemiballismus and hemidystonia (6%) and excessive daytime sleeping with or without hyperphagia (6%). The least common findings included paroxysmal attacks (4%), seizures (4%) and psychiatric manifestations, i.e. acute psychotic reaction (2%), hearing loss (2%), cerebellar syndrome (2%), optic neuropathy (1%), hemianopia (1%) and aphasia (one case).

CSF.

CSF examination was performed in 81 cases in this group. Thirty-two were completely normal. Forty-nine showed pleocytosis and/or elevated protein content. Of the 49 CSFs with a high protein or cell count, 54% had either neutrophilic predominance or both neutrophils and lymphocytes, whereas 46% showed lymphocytic predominance. Median lymphocyte count per mm3 was 30 (mean 63, range 0–485) and median neutrophil count per mm3 was 10 (mean 98, range 0–1100), and median protein level was 60 mg/dl (maximum 150 mg/dl). When evaluated, the immunoglobulin G (IgG) index was elevated in 73% of cases whereas oligoclonal IgG bands were present in 16%. However, one or two persisting bands were found in these latter patients. None of the patients showed more than two bands.

Secondary or non-parenchymal CNS involvement (n = 38; 30 males, 8 females)

Onset.

Neurological onset was with an attack in all cases. In the cases with dural sinus thrombosis, the neurological picture that ensued over a few weeks showed a subacute evolution. However, in cases with cerebral arterial involvement a stroke-like onset was seen within hours. Thirty-one cases had one attack, four had two attacks and three had three or more attacks.

Predominant syndrome.

In this group, 34 cases had the syndrome of intracranial hypertension (IH) due to dural sinus thrombosis (20 cases) and superior vena cava occlusion (VCSS) (one case). In five of the cases with IH the aetiology was undetermined despite adequate radiological studies, and the remaining eight cases did not have MRI or angiographic evaluation. One case had aseptic meningitis with papilloedema, one patient had cranial compressive neuropathy due to an external carotid artery aneurysm, one patient had vertebral artery (VA) dissection, and another patient had vasculitic deep middle cerebral artery (MCA) occlusion.

Clinical signs and symptoms.

The most common finding was papilloedema (89%), followed by unilateral or bilateral VIth nerve palsy (57%). 95% of cases had headache and 16% had high fever during the acute attack. Pyramidal signs, which were mostly unilateral, were present in 24% of the cases. Four patients had hemiparesis, three had behavioural changes, two had hemihypaesthesia, two had seizures and one patient each had sphincter disturbance, Wallenberg syndrome and compressive cranial neuropathy (IX, X, XI, XII) due to an external carotid artery aneurysm.

CSF.

CSF examination was performed in 27 cases. Four were completely normal. Twenty cases were normal except for high pressure (these were all in the cases with the syndrome of intracranial hypertension), and one showed subarachnoidal haemorrhage (the case with VA dissection). Two showed neutrophilic and lymphocytic pleocytosis, and elevated protein content (the cases with aseptic meningitis + papilloedema and vasculitic deep MCA occlusion). When evaluated, none showed an elevated IgG index or oligoclonal IgG bands.

Neuropsychological testing

Neuropsychological testing was performed in 74 of the 200 cases. Among these cases, 66 had parenchymal CNS involvement and eight had non-parenchymal CNS involvement. Only nine patients had normal neuropsychological tests (five cases with IH and four cases with silent neurological involvement), whereas all the remaining patients showed some degree of impairment. The most commonly and most severely affected function was memory processes. Memory disturbance was present in all cases with abnormal cognitive status, and in 70% of these it was moderate to severely impaired. Learning and recall processes were almost equally impaired in both verbal and visual modalities. However, memory scores substantially improved with recognition. Attention deficit was the second most common abnormality (60%), but in 55% of these it was mildly impaired. Frontal lobe functions were impaired in 52% of the patients. Other cognitive functions were relatively spared. Visuospatial dysfunction was detected in 9% of the patients and was mostly mild; abstraction was impaired in 8% and was also mild; one patient showed a mild naming difficulty. Orientation, language, arithmetic skills and directed spatial attention functions were all normal. When longitudinal follow-up of the neuropsychological performance was evaluated, it was seen that these findings were not correlated with any acute attack, and they usually tended to worsen with time irrespective of the attacks (Öktem-Tanör et al., 1999). The details of neuropsychological evaluation will be presented elsewhere.

Neuroradiology

Cranial CT scans were performed in 88 patients, and 101 patients had at least one MRI scan. Only 27 of the patients had abnormal CT scans (31%), whereas 71 patients had abnormal MRI scans (70%) (χ2 = 27, P < 0.0001). The most common lesion type during an acute attack (Fig. 1) was located in the basal ganglion region or in the brainstem, extending to the diencephalic structures (34 patients); in about one-third of the cases the lesion was bilateral. In some cases a mass effect was noted, and the centre of these lesions showed contrast enhancement. The next most common lesion type was small scattered lesions located in the basal ganglion region, the brainstem or the internal capsule (20 patients). Most of these patients were investigated a few months after an acute attack. In another 16 patients there were small scattered hemispheric white matter lesions, either periventricular or apart from the ventricles. Ten of the 101 MRIs belonged to the patients who showed only pyramidal findings at clinical examination, and all were normal. In 13 patients with IH, occlusion of dural sinuses was noted. No cortical, subcortical or cerebellar lesions were evident. In 12 chronic cases, atrophy of the brainstem and enlargement of the third ventricle disproportionate to the lateral ventricles or cortical sulci were noted. The details of cranial MRI findings will be presented elsewhere.

Fig. 1

Coronal T2-weighted MRI image of a patient examined during an acute brainstem attack shows the extensive lesion involving the right basal ganglion region, diencephalon and bilateral brainstem. This image is highly consistent with the available data on neuropathology.

Spinal MRI was performed in 12 patients. Two of these showed segmental enlargement of the spinal cord. Others were normal. Myelograms were performed in another seven patients, and all were normal.

Cerebral angiography was performed in 28 patients. Sixteen were normal. Among the remaining 12 patients, one showed an external carotid aneurysm, another showed vasculitis of MCA and another showed a VA dissection; all the remaining cases who had IH showed dural sinus thrombosis.

Evoked potentials

Evoked potential studies were performed in 40 cases. Visual evoked potentials were abnormal in five cases, who also showed severe ocular involvement. Brainstem auditory evoked potentials were abnormal in another five cases. Somatosensory evoked potentials were abnormal in seven cases, two of whom had silent neurological involvement clinically. Motor evoked potentials were abnormal in 16 cases, four of whom had silent neurological involvement clinically.

Clinical course and factors affecting prognosis

Within the follow-up period, among the 186 cases who provided sufficient information concerning the course of the disease, 77 cases had a course with at least one attack and remission(s). However, one-third of these had severe sequelae. Another 53 patients also had attack(s) but they showed secondary progression afterwards. Seventeen cases did not have any initial attacks; they had slowly but progressively evolving complaints. Six of these had an attack later in the follow-up period. The remaining 39 cases had silent neurological involvement; however, five of these cases later had an attack and remission. Age at onset of neurological involvement did not differ significantly between the groups without a progressive course and with a primary or secondary progressive course.

At admittance (see Patients and methods), 162 patients were independent, 21 patients were physically dependent, eight patients were mentally dependent and nine patients were both physically and mentally dependent on another person. Among the cases who were dependent, all but two belonged to the parenchymal CNS involvement group.

Among the cases included in the prognosis evaluation (110 cases, comprising 103 cases with ≥3 years of neurological involvement and seven deceased patients), 84 patients were initially independent, 14 patients were physically dependent, three patients were mentally dependent and nine patients were both physically and mentally dependent on another person. In this group the disability status of the patients at their last visit was independent in 60 cases, physically dependent in 12 cases, mentally dependent in eight cases, physically and mentally dependent in eight cases and deceased in 22 cases (a total of 50 cases were dependent or dead).

Prognostic factors were evaluated in 110 cases who had ≥3 years of neurological involvement or who had died within the first 3 years. Figure 2 shows that the median time to become either dead or dependent was 115.7 months. When the final outcome measure for poor prognosis was taken as either dependence or death, univariate analysis indicated that the following factors were associated with a statistically significant (P < 0.01) effect on this outcome (Table 4): parenchymal involvement in general (Fig. 3) and among this group `brainstem +' type involvement among other types of parenchymal involvement; CSF pleocytosis and/or high protein content at the acute stage (Fig. 4A and B). Being dependent at admittance, as well as a primary or secondary progressive course, was also associated with a statistically significant (P < 0.01) effect. On the other hand, the number of attacks (if two or more) and relapse during steroid dose tapering showed a weaker association with a poor prognosis (P < 0.05). The presence of certain neurological findings, such as cognitive impairment, sphincter disturbance, pyramidocerebellar findings and brainstem signs, was associated with a poor outcome (P < 0.01). A weaker association (P < 0.05) was found with pyramidal findings, fever and meningeal signs. Factors such as sex, the presence of other systemic manifestations of Behçet's disease and age at onset did not have any influence.

View this table:
Table 4

Factors that significantly affected prognosis in cases with >3 years of neurological involvement when the patients were compared according to their final disability status (n =110)

Good prognosisPoor prognosis
Final disability status: independentFinal disability status: dependent/dead
Normal CSFAbnormal CSF
(elevated protein/cell count)
Secondary or non-parenchymal involvementParenchymal involvement
(especially `brainstem +' type)
Number of attacks <2Number of attacks ≥ 2
Independent at admittanceDependent at admittance
Relapse during steroid tapering
Progressive course (primary or secondary)
Fig. 2

Event-free survival Kaplan–Meier curve of the whole group with ≥3-year duration of neurological involvement (n = 110). `Event' is defined as being either dependent on another person or death. Thus, `event-free' means being independent.

Fig. 3

Event-free survival Kaplan–Meier curve of the whole group with ≥3-year duration of neurological involvement (n = 110), according to the type of CNS involvement (parenchymal versus non-parenchymal). `Event' and `event-free' are defined in the legend of Fig. 2.

Fig. 4

Event-free survival Kaplan–Meier curves according to CSF findings (normal CSF versus presence of pleocytosis and/or elevated protein level). (A) Data for the whole group with ≥3-year duration of neurological involvement who had CSF examination (n = 71). (B) Data for the group with ≥3-year duration of parenchymal CNS involvement who had CSF examination (n = 55). `Event' and `event-free' are defined in the legend of Fig. 2.

In the whole group, 22 patients (20 male, two female) were deceased. The median period until death was 12.5 years after onset of Behçet's disease and 4 years after neurological onset. Of these 22 deaths, 18 occurred in the first 5 years of neurological involvement. Eight patients died during a severely secondary progressive course, two cases died within a progressive course but during a definite new bulbar attack, one case died of an opportunistic pulmonary infection, one case died as a complication of renal amyloidosis, three cases died during a relatively stable neurological stage due to acute respiratory distress, one patient died of acute myeloid leukaemia several years after a stable course, one patient (the only patient who had IH) committed suicide and the remaining five cases died with unknown conditions away from our medical attention. Unfortunately post-mortem examination was not available in any of our cases.

Evaluation of treatment choices is beyond the scope of this paper. However, in general terms we can say that the patients were treated as their condition required: 157 patients received corticosteroids (usually intravenous high-dose pulsed methylprednisolone at the time of the acute attack and oral maintenance thereafter), 52 patients also received cyclophosphamide (usually pulsed intravenously at the time of the acute attack and by oral maintenance administration thereafter), 63 patients received azatioprine as maintenance treatment. Cyclosporin A, chlorambucil and methotrexate were used rarely (11, five and two patients, respectively).

Discussion

Since the first description of neurological involvement in Behçet's disease in 1941 (Berlin, 1944), innumerable single cases and series of neuro-Behçet disease have been reported. In most of these reports, certain features of the disease have been hinted at. Here we present the largest series, in an effort to provide a definitive clinical picture of neurological involvement in Behçet's disease and to define factors that affect prognosis.

The current study does not provide information about the prevalence of neurological involvement in Behçet's disease, though in a previous study from our department it was found to be 5.3% (Serdaroglu et al., 1989). It has been a common mistake to include cases with only headache or other vague neurological symptoms when reporting cases with neuro-Behçet disease. To avoid confusion, cases for whom there was no evidence of neurological involvement were not included in the present study. Isolated cases of peripheral neuropathy or myopathy have been reported in Behçet's disease (Namer et al., 1987; Worthmann et al., 1996); however, in our series there are no cases with peripheral nerve or muscle involvement. A few such cases were excluded because they had a more common explanation or lacked sufficient diagnostic work-up.

With all the accumulated data it is now clear that it is necessary to handle parenchymal CNS involvement and secondary or non-parenchymal CNS involvement in Behçet's disease separately. Secondary or non-parenchymal CNS involvement is seen in ~20% of our cases. The majority of these cases have the syndrome of raised intracranial pressure, which is mostly due to a dural sinus thrombosis (or, in one case, to VCSS). In our series, non-parenchymal CNS involvement is prominently associated with a better prognosis than parenchymal CNS involvement, which supports the previous opinion that different mechanisms might be in control in the pathogenesis of these conditions. Such a good prognosis has been noted before, along with the suggestion that these cases should be called vasculo-Behçet rather than neuro-Behçet (Wechsler et al., 1992; Akman-Demir et al., 1996a; Siva et al., 1997).

Our data show that neuro-Behçet's disease (parenchymal CNS involvement) has a characteristic clinical picture. The patient is frequently a male. Onset is usually with an attack of hemiparesis evolving over a few days, with headache, gradual behavioural changes and/or sphincter disturbance. Bilateral pyramidal findings are usually present. Less frequently the patient may show brainstem signs, pyramidocerebellar syndrome or sensory findings. It should be noted that sensory symptoms are very uncommon. Fever may accompany an attack. Although the spinal cord is commonly found to be involved in autopsy series of Behçet's disease (Rubinstein et al., 1963), paraparesis is a rather less common finding. Again, despite the fact that CSF pleocytosis is common, meningeal signs are also uncommon.

Autopsy studies show that in neuro-Behçet's disease, besides a widespread but low-grade chronic lymphocytic meningoencephalitis with widespread perivenular neutrophilic or lymphocytic and plasmocytic cuffing, multifocal necrotic foci are seen; these tend to accumulate mostly in the brainstem and basal ganglion region and to a lesser extent in the spinal cord, irrelevant to the presenting clinical picture (Rubinstein et al., 1963; Sugihara et al., 1969; Hadfeld et al., 1997). In our patients we occasionally encountered hemichorea or hemiballismus, indicating basal ganglion dysfunction, and excessive daytime sleeping with or without hyperphagia, which might be indicative of hypothalamic dysfunction. On the other hand, focal necrotic lesions are very sparse in the cerebral cortex and are almost never seen in the cerebellum (Rubinstein et al., 1963; Kawakita et al., 1967; Lakhanpal et al., 1985). This is also compatible with our observation that higher cortical functions, such as aphasia, agnosia and apraxia, are very rare, as is isolated cerebellar syndrome. In 40–50% of the cases demyelination of the corticospinal tract is seen; however, focal areas of demyelination are less common (Kawakita et al., 1967; Lakhanpal et al., 1985). Optic nerves are known to be involved in Behçet's disease through autopsy or MRI studies (Rubinstein et al., 1963; Morissey et al., 1993). However, in our series we have a few cases with related clinical manifestations. Underdiagnosis is a possibility since many of the cases already have visual impairment due to uveitis.

The clinical and neuroradiological correlates of these pathological findings are quite clearly seen in our series and other, previously reported neuro-Behçet series (Wolf et al., 1965; Banna and el-Ramahi, 1991; Wechsler et al., 1993). Although the predominant clinical picture points to a certain area in the neural parenchyma, most of the cases show widespread involvement of the CNS. It should be noted that `pure' aseptic meningitis is uncommon in Behçet's disease despite common belief (Adams et al., 1997). In our series we have only one such case. Rather, the clinical picture is a widespread meningoencephalitis.

The differential diagnosis of neuro-Behçet's disease may include many diseases of the CNS. In previous years multiple sclerosis was one of the leading misdiagnoses (Totsuka and Midorikawa, 1972; Motomura et al., 1980). However, as our data show, clinical findings are distinctive in typical neuro-Behçet cases. On the other hand, symptoms which are considered to be typical of multiple sclerosis, such as paroxysmal attacks, can occasionally be observed in neuro-Behçet cases (Akman-Demir et al., 1995). On the other hand, MRI findings are also discriminative in most of the cases where the major lesion is located in the brainstem–diencephalon–basal ganglion region (Wechsler et al., 1993; Gerber et al., 1996; Akman-Demir et al., 1998). In a masked evaluation, such lesions were found to be highly suggestive of Behçet's disease (Çoban et al., 1999). However, the predominant lesion may be in the periventricular white matter (Miller et al., 1987; Morissey et al., 1993), in which case it will be difficult to discriminate from multiple sclerosis (Çoban et al., 1996, 1999). In such cases CSF pleocytosis with polymorphonuclear predominance and the absence of more than two oligoclonal IgG bands may indicate neuro-Behçet's disease (Saruhan-Direskeneli et al., 1996). The rarity of oligoclonal band positivity has been noted before (McLean et al., 1995). Other than multiple sclerosis, in certain cases of CNS infection—especially when there is CSF pleocytosis and fever—cerebrovascular disease, brain tumours and compressive myelopathy should be considered in the differential diagnosis of neuro-Behçet's disease.

Since the earliest reports it has been known that neuro-Behçet's disease has a high mortality rate, and in the 1960s this was estimated to be ~25% within the first year after neurological onset (Wolf et al., 1965). Our study is unable to provide an exact mortality rate, but we can say that it is at least 11% in a median follow-up period of 5 years. In two other neuro-Behçet series mortality rate is ~10% (Bohlega et al., 1997; Siva et al., 1998). This is much higher than the 10-year mortality rate of a general Behçet series from Turkey, which is ~5% (Yazici et al., 1996). In a much smaller follow-up series we have shown a 20% mortality rate within a period of 7 years in cases with neurological involvement (Akman-Demir et al., 1996b).

It should be noted that certain factors influence the course and prognosis of neuro-Behçet cases. In our view, the most important of these is the correlation between the acute-stage CSF findings and the clinical course. In a previous, smaller study we noted such a relationship (Akman-Demir et al., 1996b), and in the present study we verified that normal CSF at the acute stage is associated with a better prognosis, i.e. a stable course and less disability, while high cellular and/or protein content is significantly associated with a worse prognosis. This should be kept in mind when initiating treatment at the acute stage when making a decision about the addition of immunosuppressants to corticosteroid treatment. Other associations with a poor prognosis, such as `brainstem +' type involvement and a progressive course, are less surprising.

In our series, nearly 20% of the patients had no neurological complaints. They were detected at routine examination, and they can therefore be said to have had silent neurological involvement. These cases had mostly pyramidal signs with no localized findings. Such cases were also mentioned by others (Ahmed et al., 1993; Parisi et al., 1996). The clinical importance of this type of presentation is not yet well known. In our series, almost one-fifth of these patients later developed an attack. Although our impression is that these cases have a mild course, they may well represent earlier stages of primary progressive cases. This must be verified in a prospective study.

Here we should discuss the problem of patients who do not fulfil the diagnostic criteria for Behçet's disease (International Study Group for Behçet's Disease, 1990), but show `typical' features of neuro-Behçet's disease. Hentati and colleagues suggest the terms `probable neuro-Behçet' and `possible neuro-Behçet' for such cases (Hentati et al., 1993). This is a reasonable approach, since such cases may be encountered during daily clinical practice, especially in countries where Behçet's disease is highly prevalent. However, since there is no consensus on the description of these terms, we did not include such cases in the present study. In our series, neurological onset before any other symptom of Behçet's disease was seen in 3% of the cases. Onset concomitant with other symptoms of Behçet's disease was seen in 7.5% of all patients. Therefore, the diagnosis of possible or probable neuro-Behçet disease must be considered in such cases if no other cause can be established for the neurological disease.

In conclusion, neuro-Behçet's disease is commonly a motor–mental disorder with a predilection for the brainstem and the basal ganglion region, as suggested by earlier neuropathology studies and neuroradiological data. At the time of a neurological attack, typical neuroradiological findings have been defined (Çoban et al., 1999). Since no direct infectious cause has been shown, we can state that it is an aseptic meningoencephalitis. A polymorphonuclear pleocytosis and/or the absence of IgG oligoclonal bands are suggestive of neuro-Behçet's disease. The presence of worse prognostic factors should be considered when treatment is initiated. Abnormal CSF and parenchymal involvement, especially of the `brainstem +' type, justify more aggressive treatment.

Acknowledgments

*Other members of the Neuro-Behçet Study Group include: Betül Baykan, Hakan Gürvit, Mehdi Shehu, Sara Bahar, Oguzhan Çoban, Mefkure Eraksoy, Öget Öktem, Ayfer Tumaç, Jale Yazici, Emre Öge, Ari Boyaciyan, Coskun Özdemir and Edip Aktin from the Department of Neurology, Istanbul Faculty of Medicine; Güher Saruhan-Direskeneli from the Department of Physiology, Istanbul Faculty of Medicine; Ahmet Gül, Meral Koniçe and Nihat Dilsen from the Department of Rheumatology, Istanbul Faculty of Medicine; Sabahattin Yurdakul and Hasan Yazici, from the Department of Rheumatology, Cerrahpasa Faculty of Medicine; Gülsevim Azizlerli from the Department of Dermatology, Istanbul Faculty of Medicine. Parts of this work were presented at the Seventh International Conference on Behçet's Disease, Tunis, October 11, 1996 and at the Eighth International Conference on Behçet's Disease, Reggio Emilia, October 9, 1998.

Footnotes

  • * For other members of the Neuro-Behçet Study Group see Acknowledgements

References

View Abstract