Systemic lupus erythematosus (SLE) is not an uncommon condition. Most neurologists are well aware that it can cause a wide range of neurological complications, and SLE almost invariably appears on ‘differential diagnosis’ lists in cases of clinical uncertainty. However, the precise nature of the manifestations of SLE in the central and peripheral nervous systems is perhaps less widely understood, and misperceptions about phenomenology and treatment are common. Here we survey some of the main primary neurological complications of SLE—‘neurolupus’—while acknowledging that secondary problems, either iatrogenic or relating to other consequences of SLE (eg, hypertensive CNS disease, for example, secondary to renal lupus) are neither less serious nor less treatable.
Statistics from Altmetric.com
Systemic lupus erythematosus (SLE) is an autoimmune disease with diverse clinical manifestations ranging from trivial and ephemeral to life threatening. Improved therapeutic options in recent years have enhanced survival rates in the face of a rising incidence over the past four decades (the last probably in part due to better recognition). The current estimated prevalence of SLE ranges from 6 to 150 per 100 000, and depends on sex and racial origin— there is a 5–10 times higher risk in females, African–Caribbeans and Asians.1 The diagnosis is usually based on the American College of Rheumatology (ACR) criteria (table 1) which require at least four of the 11 listed features at some time in the course of the illness.2
Involvement of the nervous system—cognitive and psychiatric, central and peripheral nervous, all inclusively referred to as ‘neurolupus’ in this brief account—represents a much dreaded complication, and is common.3 Precisely how common has proved difficult if not impossible to establish. Published studies give figures ranging (remarkably) from 14% to over 90% of SLE patients developing neurological or psychiatric symptoms at some time during their illness, the unhelpfully large range a result of studying fundamentally different populations and of substantial differences in the definition and criteria used to designate nervous system involvement.4,–,6 Many authorities, possibly depending more on resigned guesswork than sound epidemiology, suggest 50% may be about right. Despite its gravity, however, SLE is potentially treatable, making it important for neurologists to be aware of the spectrum of neurolupus and its management.
Pathology and Pathogenesis
In order to understand neurolupus and to rationalise investigations and treatment, it is useful to have a basic understanding of the underlying disease mechanisms.7 The principal neuropathological changes are those of infarction in the territory of small vessels, particularly in the cerebral cortex and brainstem (table 2). Vasculitis (true inflammatory infiltrate and destructive change within the blood vessel wall) is rare in neurolupus. There is extravasation of fibrin and red blood cells, together with endothelial cell proliferation, hypertrophy and the appearance of fibrin thrombi9 (figure 2). These non-inflammatory proliferative changes appear to be responsible for the numerous areas of microinfarction. Less commonly, macroscopic infarction or haemorrhage occurs, the former occasionally due to embolism from Libman–Sacks endocarditis, or to the consequences of antiphospholipid antibodies (APAs).
A common suggestion is that focal manifestations, including stroke, are predominantly caused by vascular occlusion, while diffuse manifestations, such as psychiatric features, may result from pathogenic antineuronal antibodies. Indeed, a hallmark of SLE is autoantibody production (table 3, figure 3).
In principle, in neurolupus, antibodies could mediate neurological injury by reacting against neuronal, astroglial or endothelial cells.7 Changes could also be induced in the cerebral vasculature, either through immune complex deposition or effects on the coagulation system leading to infarction. These proposals are not mutually exclusive. In addition, well recognised abnormalities of cytokines, complement components, immune related genes and hormones could all contribute to lupus related processes. Much recent interest has focused on apoptosis (programmed cell death) in SLE. Antibodies to C1q complement protein are common in SLE, and can cause a functional deficiency of C1q and so defective clearance of apoptotic cells leading to an increase in circulating nucleosomes (essentially circulating DNA fragmentation products of apoptosis) and so to antinucleosome antibody production.7 These theories could help explain some of the precipitants of disease activity; (for example, sunlight and viral infections) may increase the number of apoptotic cells in the skin and blood with release of intracellular antigens leading to increased autoantibody generation. These diverse mechanisms probably overlap to produce the spectrum of phenotypes typical of SLE.
The Clinical Manifestations of Neurolupus
Headaches are common, but chronic tension-type headaches are no more frequent than in the general population. The incidence of migraine has been suggested to be higher. However, a recent meta-analysis has been extremely helpful in exploring the relationship between headache and SLE, concluding that “the prevalence of all headache types, including migraine, was not different from controls” with “insufficient evidence … for the concept of ‘lupus headache’”.15
‘Idiopathic’ intracranial hypertension has been described in lupus, sometimes as an initial feature, with the typical findings of bilateral papilloedema and raised CSF opening pressure without focal neurological signs.
Seizures are a well recognised complication,16 occurring in 14–25% of SLE patients compared with 0.5–1% in the general population. They are usually generalised but partial episodes also occur. Although accompanying systemic and other CNS features are usually present, seizures may be one of the earliest manifestations of neuropsychiatric involvement, sometimes occurring several years prior to generalised SLE, understandably leading to the erroneous diagnosis of an isolated epilepsy. Several mechanisms may cause seizures in SLE but CSF pleocytosis in such cases raises the possibility of a low grade lupus related encephalitis. Furthermore, the common finding of cerebral atrophy in SLE may also predispose to seizures.
Stroke and recurrent transient ischaemic attacks occur in 5–20% of SLE patients, at a mean age of 37 years (range 26–50 years), from our own study,17possibly leading to multiinfarct dementia in some cases. Intracranial venous thrombosis is another complication. There is evidence to support an APA and lupus anticoagulant induced coagulopathy but the alternative possibilities of embolism from Libman–Sacks endocarditis or hypertension from nephritis must not be overlooked.
The antiphospholipid antibody syndrome is defined as the association of APAs with arterial or venous thrombosis, recurrent fetal loss, thrombocytopenia or neurological disorders, such as stroke, transient ischaemic attacks, transverse myelopathy, chorea, seizures and migrainous headache.18, 19 It was first described in patients with SLE (secondary antiphospholipid syndrome) but it may also occur in the absence of any other disorder (primary antiphospholipid syndrome). A devastating malignant form with multiorgan involvement including the CNS, catastrophic antiphospholipid syndrome, is well recognised.20 A detailed discussion of this group of disorders closely related to neurolupus is beyond the scope of this brief review.
Psychiatric and cognitive disturbances21, 22 occur in 30–75% of SLE patients, the wide range reflecting large variation in diagnostic criteria for such disorders. There is no single characteristic clinical picture although three relatively distinct syndromes are recognised.
‘Pure’ behavioural or psychiatric illness without clouding of consciousness.
In the first category, affective disorders such as depression and anxiety are particularly common, and while there is evidence that these could be a direct manifestation of SLE, they may be a secondary response to a life altering chronic illness. A ‘lupus psychosis’ is also described23 (although one should be careful to differentiate it from iatrogenic steroid induced psychosis). Delusions, visual and auditory hallucinations, catatonia and conversion disorders are all recognised.
Cognitive impairment24 (particularly objective memory testing abnormalities and deficits in attention and visuospatial function) are frequent. Interestingly, impairment of neurocognitive function may be observed in patients, even when their illness is apparently quiescent. Various case series of neurolupus in older people have emphasised subacute confusional state, dementia and depression, and suggested that neurolupus may be more common in this age group than previously assumed.17 Underlying mechanisms such as temporary autoantibody effects or the indirect effects of depression mean that cognitive dysfunction in SLE is potentially reversible— although this does not apply to multi-infarct dementia.
Aseptic meningitis—acute, chronic or recurrent—is rare although pathology studies demonstrate meningeal inflammation in almost one-fifth of cases. It is seldom an initial feature of SLE but tends to occur early in the course of the disease and may herald further CNS problems such as transverse myelitis and strokes. SLE should therefore be included in the differential diagnosis of recurrent aseptic meningitis. Aseptic meningitis in SLE can also be associated with non-steroidal anti-inflammatory drugs. Fortunately, withdrawal of the offending drug and steroid therapy invariably lead to recovery.
Movement disorders such as chorea are often regarded as a classical neurological complication of SLE but in fact occur in less than 4% of SLE patients. They are more likely to appear during an acute flare of the disease, particularly in women under the age of 30 years who tend to develop generalised or hemichorea. This is usually subtle and transient but may be more prolonged, recurrent or permanent in some. Chorea of SLE is an important differential diagnosis of rheumatic fever in children. Hemiballismus due to infarction of the subthalamic nucleus, ataxia due to cerebellar and brainstem involvement, myoclonus, torticollis and blepharospasm have all been documented. Tremor is recognised in SLE but parkinsonism is rare; the few case reports in patients of varying ages have generally had severe CNS involvement with little or no response to antiparkinsonian medication but good improvement with steroids and immunosuppressants.
Myelopathy is one of the most disabling complications of SLE but fortunately is unusual,25, 26 with a prevalence of 1–5% of SLE patients. Despite this infrequency, it has a tendency for early involvement: in 39% of neurolupus cases with myelopathy, the spinal disorder occurs as the initial presentation of SLE, while in a further 42% spinal involvement occurs within the first 5 years of SLE diagnosis. Lupus myelopathy is usually acute or subacute, with a raised CSF protein and lymphocyte count (rarely the CSF is normal). MRI with gadolinium enhancement shows changes in the affected region of the cord, often extensive in length in the cervical to mid-thoracic region, usually with cord swelling. Interestingly, a link with antiaquaporin antibodies and neuromyelitis optica (NMO) has emerged recently.27
Cranial neuropathy is relatively uncommon and often transient. An ‘autoimmune optic neuropathy’ is well described, and again there is an emerging link with NMO and its associated autoantibody.27 Trigeminal sensory neuropathy or typical trigeminal neuralgia may occur as an isolated neurological manifestation in SLE or as part of a mononeuritis multiplex.
Demyelination in SLE? On the relationship between demyelinating disease and neurolupus, the pendulum has swung most engagingly over the past few decades. In 1972, Fulford et al coined the term ‘lupoid sclerosis’ to describe clinical similarities, and a possible underlying aetiopathogenetic link, between SLE and demyelinating syndromes.28 Optic neuritis, brainstem and cerebellar syndromes, subacute myelopathy and other transient neurological deficits in the context of SLE may all mimic multiple sclerosis clinically and radiologically. However, pathological studies of individuals with CNS lupus confirmed that this is a distinct disorder, with no evidence of primary demyelination9; the consensus became that the phenotypic similarities did not reflect common mechanisms.7 More recently still, however, the occurrence of optico-spinal syndromes clinically resembling NMO, combined with the important observation of seropositivity for NMO antibodies, has re-emphasised the possibility of a shared immunological basis for some neurolupus phenotypes and the NMO subtype of demyelinating disease.27
Other ocular structures are commonly involved in SLE, requiring careful ophthalmological evaluation. Furthermore, the clinician needs to assess whether the eye condition is caused directly by SLE, a complication of SLE, or by toxicity of therapy.
Retinopathy is frequently observed with cotton wool exudates (indicative of local retinal ischaemia) and haemorrhages, and less frequently, perivascular sheathing and fibrosis.
Abnormal eye movements and ptosis (often unilateral) are occasionally reported as a recurrent feature in a number of patients. Proposed mechanisms include focal or widespread brainstem disease, or cranial nerve vasculitis.
Choroidopathy is probably more common than generally appreciated and is often an indicator of severe underlying systemic disease but uveitis is uncommon.
Peripheral nervous system involvement.29 Although less common than CNS involvement, peripheral neurological complications do occur in SLE:
Acute inflammatory demyelinating poly- neuropathy (Guillain–Barré syndrome).
Chronic inflammatory polyneuropathy.30
There is a suggested association with myas- thenia gravis31 (which should be particularly considered as a differential diagnosis of some of the possible ocular manifestations described above), the Lambert–Eaton syndrome and following thymectomy.
While not considered in detail here, neuropsychiatric complications resulting from factors other than the primary disease process should not be neglected in practice.
Intercurrent infections and immunosup - pression by drug therapy may be major causes of morbidity and mortality, making it essential to exclude conventional and opportunistic organisms.32
Other adverse effects of drugs are com- mon; steroid therapy in particular can cause psychosis and hypertension.
Chronic renal failure may increase blood pressure and lead to metabolic derangements with further potential for adverse neuropsychiatric effects.
Risk factors for CNS involvement and outcome in SLE
There appear to be several risk factors for CNS involvement and a poor prognosis.33Individuals with prior neuropsychiatric involvement are at greater risk than other SLE patients of developing (further) neurological complications. There is also an increased risk with the antiphospholipid syndrome as well as its features (in particular, arterial thrombosis and cutaneous vasculitic lesions), and with thrombocytopenia, positive antinuclear antibodies (ANA), anti-SS-B/La and low serum levels of C3 and C4 complement components. Patients with this profile may require closer observation and are potential candidates for studying pre-emptive interventions. Some—but not all—authors believe that articular involvement and discoid rash may be protective against the development of neurolupus features.
Making the Diagnosis of Neurolupus
The many clinical manifestations of SLE can make the diagnosis difficult—after all, there are 120 different combinations of ‘four out of 11 features’ of the ACR diagnostic criteria (table 1). Furthermore, these are permitted at ‘some time in the course of the illness’ and one should therefore not exclude the possibility of SLE, even if the criteria are not satisfied at the outset. The next stage involves confirming the presence of any of the neuropsychiatric features, as outlined in the ACR list (table 4). Such features may be very early or even the first presentation of SLE, posing a significant diagnostic challenge.17 Recommended investigations for neurolupus are listed in table 5.
Blood tests and serology
There is no specific diagnostic blood test for neurolupus, but in conjunction with relevant clinical features of the disease itself, various blood tests are nonetheless helpful.
Routine biochemistry may detect associated renal impairment or electrolyte disturbances, either from SLE or iatrogenic drugs.
A full blood count may show anaemia, leucopenia or thrombocytopenia.
The erythrocyte sedimentation rate is a sensitive but non-specific indicator of disease activity in SLE; C reactive protein (CRP) has a shorter half-life and rapidly reflects acute inflammation. Classically (but not reliably) in active SLE, the erythrocyte sedimentation rate is raised and the CRP is low. Therefore, a very high CRP may distinguish between bacterial infection and active SLE. Low C3 and C4 levels may be helpful but SLE can be active without causing any of the above changes.
More diagnostically discerning is the assay of autoantibodies (table 3). These are usually polyclonal and often directed against multiple targets.
As the name implies, ANA are antibodies directed against Although persistent nuclear components. ‘ANA negative’ SLE is recognised,34 it is not clear if this truly represents a subgroup of SLE or a technical artefact. No specific titre of ANA is defined as abnormal by the ACR, but in general, high titres (>1/160, but preferably >1/300 depend- ing on the laboratory and test used) in young patients are diagnostically meaningful. One needs to consider these results in their clinical context, as positive ANA are also encountered in those who are unwell or older and even in some normal individuals.35
Further specific tests looking for anti-DNA antibodies and extractable nuclear antibodies are more defining of SLE. Anti-DNA antibodies are a reasonably specific diagnostic test but are positive in only about 70% of SLE patients, rising in active disease—they can be negative early in disease, following treatment, or in clinical remission, and are also sometimes found in normal individuals, in Sjögren’s syndrome and in those with rheumatoid arthritis (see also table 3).
APAs are seen in 16–60% of those with SLE. Although not all patients with these antibodies have the APA syndrome, IgG antibodies and β2 glycoprotein 1 in particular are recognised risk factors for thrombosis. Additional lupus anticoagulant testing is essential because this finding also predisposes to thrombosis and may occur in the absence of APAs.
Although it would be unusual for all of these antibodies to be absent in a case of neurolupus, their temporary absence does not (sadly) exclude the diagnosis.
Spinal fluid examination
There are mild and non-specific abnormalities in 30–90% of patients with neurolupus; a modestly raised white cell count (predominantly mononuclear) and protein. The CSF glucose is usually normal but can occasionally be depressed, especially in SLE related myelopathy. Oligoclonal bands are reported in 15–80% of cases (both matched and unmatched with serum); these appear to be promoted by specific cytokines in the CSF such as interleukins 6, 8 and 10, which are increased almost 10-fold in the CSF of those with neurolupus compared with nonCNS lupus. Furthermore, reduced interleukin 6 CSF activity after remission of CNS symptoms with cytotoxic therapy is recognised while resistance to treatment is associated with persistent elevation.
Although CT brain imaging is not the most sensitive marker of neurolupus, it readily detects and differentiates between infarcts and haemorrhages, with generally good correlation with stroke-like presentations. One consistent finding in chronic cases is cortical atrophy (usually perisulcal), which may be a direct result of local lupus related disease processes or secondary to microinfarcts or long term steroid therapy. MRI is far better at revealing small infarcts, focal cerebral oedema and (of course) myelitis.
Unsurprisingly, the pattern of clinical involvement is reflected in the MRI features: localised neurological signs and seizures are associated with focal MRI abnormalities more frequently than a diffuse presentation with an organic brain syndrome, psychosis, headache and/or meningism. There is however no one MRI finding or pattern that is diagnostic or specific for neurolupus.36, 37 Nonetheless, some characteristic changes have been proposed: acute lesions typically lack discreet borders, are intermediate intensity on T2 weighted images, have a lacy/filamentous pattern, are semilunate in shape and are located at the grey–white matter junction along sulci and gyri, or in the white matter with overlying grey matter hyperintensity. These lesions tend to enhance with gadolinium, so contrast studies are recommended within 24 h of the neurological event. These high intensity lesions may potentially resolve with steroids. In addition, MR sequences with fluid attenuated inver sion recovery and diffusion weighted imaging enhance the sensitivity and specificity for the diagnosis, but only partially (figure 4).
Clinical–radiological correlates are not always obvious. MRI lesions may represent a prior attack of neurolupus or may resolve completely within days in parallel with clinical improvement. To complicate the situation fur ther, a significant proportion of SLE patients with neither active nor historically identifiable neurolupus have MRI abnormalities which increase with age and with the presence of vascular risk factors. These chronic lesions tend to be small, punctate, focal white matter changes that rarely enhance with gadolinium. They are indistinguishable from age related small vessel disease changes but are probably more common than in non-SLE, although there are no population based comparative studies.
Volumetric magnetisation transfer imaging is more sensitive to structural brain damage than conventional MRI. It has proven to be particularly valuable in distinguishing between active neurolupus and residual disease, which may in turn help in establishing who to treat. It may also be useful for treatment trials and in for studying the natural history of the disease.
In practice, serious neurological lupus is not common.
Peripheral nervous system involvement is significantly less common than CNS lupus.
Targeted investigations, properly interpreted, are often diagnostic, while imaging is generally more important in excluding alternative diagnoses than confirming neurolupus.
Both ‘lupoid sclerosis’ and ‘lupus vasculitis’ are (usually) misleading terms.
There is very little evidence on which to base treatment recommendations; but in thrombotic disease, previous advice that warfarin is superior to aspirin, and that high dose warfarin is better than low, are both now seriously questioned. Cyclophosphamide and steroids should probably be used in inflammatory disease.
Functional brain imaging
Altered cerebral blood flow (CBF) in neurolupus patients has been detected by radionuclide brain scans, positron emission tomography (PET) and single photon emission computed tomography (SPECT). Both a decrease in mean CBF and an abnormal CBF regional distribution have been observed in diffuse as well as focal CNS presentations. These often normalise after resolution of the CNS symptoms and sometimes after treatment. PET and SPECT are sensitive although non-specific diagnostic imaging tools. They are finding a niche in helping to detect subclinical CNS involvement in SLE, in revealing irregularities in patients with cerebral involvement but normal MR scans’ and in helping to explore underlying pathological mechanisms (eg, the role of hypoperfusion) in certain SLE manifestations.38, 39
CT and MRI of patients with chorea in SLE (a putative antibody mediated neurological manifestation) are usually normal; here there is a good illustration of the value of functional imaging—18F deoxyglucose PET in other choreic disorders reveals characteristic glucose hypometabolism in the striatum which is not reproduced in SLE patients with chorea, supporting the possibility that chorea results from the effects of cytotoxic antibody (possibly APA) on the basal ganglia.40
EEG is regarded as a very sensitive though non-specific investigation in cerebral lupus, with abnormalities in even the most subtle or diverse of neuropsychiatric manifestations. Although any changes are often diffuse, some researchers have specified more focal abnormalities with selective involvement of the left temporal or temporolimbic regions.
The important pathological features have been described above. The character of cerebral involvement in SLE and the perceived risks of biopsy in these circumstances, however, often mean that tissue diagnosis is made only at autopsy, if at all; in practice, the greater role of biopsy is arguably in the exclusion of other disorders. Peripheral nerve biopsy is easier and safer, and so may have a useful role in investigating neuropathy in SLE.
The Treatment of Neurolupus
Although neurolupus is often serious and sometimes fatal, it is potentially treatable. Unfortunately, there is an extraordinary lack of large randomised trials and so retrospective studies and observations provide the main evidence. Notwithstanding the array of neurological disorders in SLE and their multifactorial aetiology, therapeutic efforts largely fall into three categories—stroke prevention, immunosuppression and symptomatic treatment (the last we will not discuss).
There is evidence that APAs, β2GP1 and the lupus anticoagulant are significant independent risk factors for stroke, and most authors do not doubt the use of antithrombotic therapy in patients with these antibodies and thromboembolic events. However, the evidence base for therapy is by no means perfect. Various guidelines commonly propose that warfarin is superior to aspirin, and high dose warfarin best of all, but recent trials and systematic reviews suggest otherwise. While further studies are required, an emerging consensus suggests that warfarin at doses aiming for a higher international normalised ratio range than usual confers significantly greater risks but no benefits in patients with APAs41; and in fact, warfarin does not appear to be superior to aspirin in the prevention of arterial thromboembolic complications in patients with a first ischaemic stroke and APAs.41,–,43 It should be stressed that this evidence largely comes from studies with primary antiphospholipid syndrome; recommendations for patients with APAs in the context of SLE are far less clear-cut. Even less consensus has emerged for primary prevention although aspirin is commonly recommended.
Corticosteroids and immunosuppressants
We are aware of only a single randomised trial assessing steroids and/or immunosuppression in neurolupus, and this included only 32 patients.44, 45 It suggested that for serious (non-stroke) neurological complications—including seizures, optic neuritis, neuropathy, coma, brainstem disease or transverse myelitis—steroids alone were inferior to steroids plus cyclophosphamide. There have, however, been many studies in lupus nephritis which clearly reach the same conclusion, and in the absence of a better evidence base, a common recommendation is therefore to start with high dose intravenous methylprednisolone, 1 g daily for 3 days (followed by oral prednisolone 60 mg/day, decreas- ing by 10 mg at weekly intervals to 10 mg/day if possible) accompanied by oral or intravenous cyclophosphamide. Oral administration is easier—2.5 mg/kg/day (lower dose of 2 mg/kg in the older people or in renal failure) for perhaps 9–12 weeks (pulsed weekly intravenous cyclophosphamide appears to differ insignificantly in efficacy from daily oral treatment and it may have fewer adverse effects). Careful monitoring of the blood count for bone marrow suppression should force a reduction in dose if there is leucopenia (total white cell count <4.0×109) or neutropenia (<2.0×109). Cyclophosphamide is associated with haemorrhagic cystitis (a complication reduced by adequate hydration and MESNA cover), a 33-fold increase in bladder cancer, other malignancies, infertility, cardiotoxicity and pulmonary fibrosis.
Conversion at around 3 months to a maintenance regimen of alternate day steroids (10–20 mg prednisolone) and substituting azathioprine (2 mg/kg/day) for methotrexate (starting dosage 7.5 mg/week, increasing or decreasing 2.5 mg/week according to response or adverse effects; maximum 20 mg/week) for cyclophosphamide can be instituted for a further minimum of 10 months, depending on response, followed by gradual withdrawal.46 Mycophenolate and ciclosporin may be as effective as methotrexate and azathioprine.47
Case reports and small series suggest intravenous immunoglobulin may be of value, and thalidomide also appears to have helped some patients. Biological agents—principally monoclonal antibodies—designed to interfere with T cell activation, T cell–B cell collaboration or to inhibit production or deposition of anti-dsDNA antibodies and complement, or cytokine activation, are all under active consideration.48, 49
This article was reviewed by Richard Davenport, Edinburgh, UK.
Competing interests None
Provenance Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.