Statistics from Altmetric.com
The last decade has seen significant advances in our understanding of the spectrum of probable antibody-mediated central nervous system disease and the autoantibodies that can help in their identification. In their recent review for this journal, Paul Gozzard and Paul Maddison detailed the wide range of rare neurological conditions associated with paraneoplastic (onconeural) antibodies, which are usually directed at intracellular antigens. Here, the authors focus on antibodies directed against mainly cell-surface neuronal antigens which are associated with central nervous system conditions that often respond to immunotherapies.
Antibodies that target the extracellular domain of neuronal cell-surface proteins and thus are likely to contribute directly to the pathology of the condition. Patients with these antibodies often do not have associated tumours and tend to respond well to immunotherapies, with tumour treatment, if appropriate. Successful treatments are often associated with paralleled reductions in antibody concentrations. Some appear to have a monophasic course but others may relapse and require ongoing immunosuppression.1
By contrast, onconeural antibodies usually target intracellular proteins and represent markers of an associated cancer but have not, in general, been shown to be pathogenic.2 Patients with these onconeural antibodies usually do not respond to immunosuppression even in addition to tumour removal. Nevertheless, because of the importance of identifying an underlying neoplasm, onconeural antibodies are often also requested when an autoimmune condition is suspected, even when a cell-surface neuronal autoantibody is found.
Although all currently recognised antibody-mediated non-paraneoplastic central nervous system (CNS) disorders are rare, numbering perhaps 5–10 per million per year (based on current results of routine antibody screening from the Oxford Clinical Service), it is important to identify these disorders so that appropriate treatments can be offered. Here, we describe the syndromes and provide guidelines to the antibody tests that are most likely to be helpful in each clinical scenario. It is important to appreciate that there are patients whose syndromes appear to be identical to the established phenotypes, but who are negative on the tests currently available. Moreover, as mentioned below, autoantibodies to specific neuronal proteins are beginning to be recognised in a proportion of patients with otherwise unexplained subacute onset of neurological symptoms, with or without MRI and cerebrospinal fluid (CSF) evidence of inflammation. In all of these patients, immunosuppressive treatments should be seriously considered, although it needs to be made clear that there have been no formal treatment trials in most of the conditions discussed here.
Limbic encephalitis (LE)
LE typically presents with acute to subacute onset of memory loss, confusion and seizures (table 1). Classically, a combination of MRI evidence of hippocampal swelling/inflammation, mediotemporal EEG abnormalities, abnormal CSF and/or immunohistological evidence of inflammation would be required for the diagnosis. However, with detection of a specific antibody, the diagnosis can be made on clinical, radiological and serological grounds alone.3 Since the previous Practical Neurology guide to LE in 2006,4 several new antibodies have been identified that are associated with LE. Ideally such patients should now be tested for serum (paired with CSF if possible) antibodies to voltage-gated potassium channel (VGKC) complex antigens (leucine-rich glioma inactivated protein 1 (LGI1), contactin-associated protein-2 (CASPR2) and contactin-2), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs), γ-aminobutyric acid-B receptors (GABABRs), glutamic acid decarboxylase (GAD), N-methyl-d-aspartate receptor (NMDAR) and for onconeural antibodies (particularly anti-Hu, anti-Ma 1/2, CV-2, amphiphysin, but most laboratories will first perform screening tests to look for all onconeural antibodies and then confirm with specific immunoblotting).2 However, to avoid cost and delay, until wide panels of the new antibodies are available routinely, it is reasonable to limit the antibody request to those that are most likely to be relevant. In table 1 we emphasise those clinical and epidemiological features that may help to distinguish between the LE patients with different antibodies.
The VGKC-complex antigens: LGI1, CASPR2 (and contactin-2)
VGKCs are found throughout the brain and are important in restoring the membrane potential during hyperpolarisation. Antibodies to the VGKC (VGKC-Abs) have been identified since the 1990s using a 125I-α-dendrotoxin-labelled radioimmunoprecipitation assay (RIPA). Dendrotoxin labelled VGKCs were extracted from rabbit brain tissue and positive results were first described in patients with acquired neuromyotonia.5 This condition includes spontaneous and continuous muscle fibre activity, painful cramps and impaired muscle relaxation; electromyography (EMG) shows doublet, triplet or multiplet single unit discharges with high intraburst frequency.6 Some patients with neuromyotonia also have autonomic dysfunction.
VGKC-Abs were then identified in the rare Morvan's syndrome, where symptoms of neuromyotonia are accompanied by florid autonomic features and CNS involvement.7 Subsequently, VGKC-Abs were identified in a form of LE8 9 and more recently in patients with faciobrachial dystonic seizures (FBDS).10
It is now clear that the VGKC-Abs detected by RIPA are not in fact directed at the VGKCs themselves (except in rare cases and perhaps in patients with low titres). They mainly bind to other components of the VGKC-complexes that are present in mammalian brain extracts, and are now better termed VGKC-complex antibodies. The main VGKC-complex proteins identified as antibody targets so far are LGI111 12 and CASPR2.11 13 It is likely that antibodies to these proteins will also be found in some patients who are not positive for VGKC-Abs by RIPA (E Becker, P Pettingill, A Vincent, unpublished), but since there may well be more VGKC-complex proteins to identify, and until more sensitive and reliable commercial antibody testing becomes available, we continue to use the RIPA for VGKC-complex antibodies, available as a commercial kit (RSR Ltd, Cardiff, UK), with assays for the specific antigens (see below) if appropriate. Figure 1 illustrates the possible localisations of the VGKC-complex proteins, and figure 2 demonstrates binding of serum antibodies to LGI1 detected using a cell-based assay.
LE with VGKC-complex antibodies
Patients with VGKC-complex-antibodies typically present with the triad of memory loss, confusion and seizures.8 9 14 They were first defined on the basis of high titres of VGKC-complex-antibodies (>400 pM). However, lower levels can be found in patients with typical LE15 as well as in around half the cases with neuromyotonia. An important and diagnostically useful feature of LE patients is serum hyponatremia (in around 60%),9 11 12 16 the CSF is usually normal and seldom with oligoclonal bands.8 16 EEG may show diffuse slowing with occasional epileptogenic foci. The MRI brain scan typically shows increased bilateral medial temporal lobe (MTL) signal,11 12 but with increased recognition it is clear that this occurs in only around 50% of patients and some have unilateral involvement.11
VGKC-complex antibodies are remarkably susceptible to immunotherapies and the antibody titres tend to fall rapidly with effective immunotherapies (eg, high dose oral corticosteroids or pulsed methylprednisolone, plasma exchange and/or intravenous immunoglobulins).8 9 11 14 The seizures usually cease and the hyponatraemia resolves. Cognitive changes often take longer to resolve but moderate to substantial functional improvements can be achieved.11 The MRI changes may resolve but hippocampal atrophy is a common observation at follow-up. Late onset of hippocampal sclerosis and ensuing temporal lobe epilepsy have been identified in adults whose MRIs suggest that they had a preceding form of LE.17
Most patients with typical LE and VGKC-complex antibodies have antibodies directed against LGI1 and usually do not have a tumour.11 12 By contrast, patients with antibodies to the VGKC-complex-protein CASPR2 commonly have neuromyotonia or Morvan's syndrome and a significant incidence of tumours particularly thymomas (often associated with previous or concurrent myasthenia gravis).11 13 However, there is some overlap and a few patients with LE do have CASPR2 antibodies.11
FBDS as an epileptic prodrome to LGI1/VGKC-complex antibodies LE
Most recently, high VGKC-complex Ab titres (238 to 8800 pM) have been found in adult patients presenting with dystonic seizures involving the arm and face. These FBDS are very brief (a few seconds) but occur at a median of 50 times a day and have been identified both prospectively and retrospectively in around 20% of VGKC-complex-antibody LE patients.10 Ninety per cent of the patients had LGI1 antibodies. Interestingly, many patients developed FBDS before the onset of amnesia or confusion. In addition, a few cases were treated with immunotherapies and never developed cognitive dysfunction. Ninety-six per cent of patients showed a greater than 20% reduction in FBDS frequency with these treatments. By contrast, their response to standard antiepileptic drugs was poor.10 Therefore, FBDS may represent an immunotherapy-responsive prodrome to VGKC-complex LGI1-antibody associated LE. Indeed, we are now aware of UK clinicians who have diagnosed and treated this syndrome prior to the antibody result without the patients developing amnesia (SRI, AV, MR Johnson, unpublished observations).
Morvan's syndrome is not diagnosed frequently. It is defined as a combination of muscle fasciculations and cramps, autonomic disturbance and CNS involvement with frequent insomnia and hallucinations.7 It can be misdiagnosed as schizophrenia until the muscle involvement is recognised.18 VGKC-complex antibodies are present in most reported cases, with a high proportion directed at CASPR2 and the patients can improve dramatically after immunotherapies.7 18 Thymomas are a commonly associated malignancy in the cases reported to date, and the patients may have co-existing, or previous histories of myasthenia gravis.11 13
LE with AMPAR antibodies
The AMPAR is a subtype of the glutamate receptor (GluR), composed of tetramers of the GluR types 1, 2, 3 or 4 and known to mediate fast excitatory neuronal transmission. Antibodies to the AMPA GluR1 and GluR2 subunits have been found in patients with LE, many of whom have thymus, breast or lung tumours, but some patients had no tumours identified.19 Unlike the monophasic VGKC-Ab LE, these patients have a tendency to relapse, and some require longer term immunosuppression with azathioprine. CSF may show leucocytosis and elevated protein.
A further report of patients with AMPAR-Ab LE noted two patients who presented with rapidly progressing abnormal behaviour similar to an acute psychosis, who had normal CSF and MRI.20 Physicians need to be aware that atypical psychosis may be linked to AMPAR-Ab and other autoantibodies (see below).
LE with GABABR antibodies
A further target for antibody-mediated LE is the GABAB1R, which mediates synaptic inhibition. Of the fifteen reported patients with these antibodies, the predominant presenting symptom was seizures of temporal lobe onset.21 Patients had increased MTL signal on MRI and CSF often showed lymphocytic pleocytosis. There was some overlap with other antibodies: a few had antibodies to GAD. Half had tumours, mostly small cell lung cancer, but most (9/15) responded to immunotherapy.21
GAD is an enzyme required to convert the excitatory neurotransmitter glutamate to the inhibitory neurotransmitter GABA. Lack of this conversion would be expected to lead to motor and CNS hyperexcitability. Autoantibodies to GAD were first described in stiff person syndrome (SPS).22 Unlike the previously discussed antibody targets, GAD is located intracellularly, but high titres of GAD antibodies appear to be markers for immune-mediated diseases including SPS, cerebellar ataxia, epilepsy and a form of LE.23 Few of these patients have tumours but many have other autoimmune diseases.15 23
There are several reports of patients with GAD-Abs and LE or epilepsy. In one study of nine patients with intractable seizures seen at an epilepsy centre, the patients with GAD-Ab LE were mostly young adult females who presented with seizures without memory loss.15 CSF was often normal although some had oligoclonal bands and MRI showed increased amygdalohippocampal signal, which was included in the selection criteria, usually bilaterally. Unlike VGKC-complex antibodies LE patients, those with GAD-Ab LE followed a chronic course, and serum antibody titres did not fall substantially following the immunotherapies used.15
Encephalitis with NMDAR antibodies
This syndrome really needs to be considered as a distinct entity from LE because of the clinical involvement of much wider brain functions and frequent lack of MRI evidence for temporal lobe swelling or inflammation (table 2).
NMDARs are ligand-gated ion channels, known to be major mediators of excitatory neurotransmission. Antibodies to the NMDAR (the main target being the NR1 subunit) were initially described in a paraneoplastic encephalitis associated with ovarian teratomata, with neurological symptoms presenting prior to tumour diagnosis (range, 1–380 weeks).24 25 The first symptoms were usually psychiatric, behavioural disturbances and seizures, with clinical worsening over 10–20 days to generate dyskinesias (orofacial grimacing, dystonic posturing and choreoathetoid movements), autonomic instability and reduced consciousness, often requiring ventilatory support.26 27 The two-stage progression of these symptoms may involve cortical and subcortical functions26 27 but this pattern may not be as common in children.28 Pleocytosis and NMDAR antibodies in the CSF, with high intrathecal antibody synthesis are very frequent25,–,27 but serum NMDAR-antibody levels are higher in absolute terms.26 MRI findings vary from normal to mild abnormalities, usually outside the MTLs, in the cerebral cortex, overlying meninges or basal ganglia.25,–,27 A recent review covers the clinical and paraclinical findings in more detail.27
With increasing clinical awareness of the syndrome, the proportion of patients with an ovarian tumour has decreased and a higher proportion of children and men are being identified with the syndrome.26 27 Paraneoplastic cases are infrequent in children26,–,28 and are very unusual in men.26 In those patients with ovarian or other tumours, the syndrome responds to immunotherapy once the tumour is removed; later removal can be associated with poor outcomes.25,–,27 Although there is no formal evidence for the effects of immunotherapies or for which treatment is most appropriate, patients do seem to benefit from aggressive immunotherapy.26 27 Patients may recover completely or be left with residual cognitive impairment and usually all have amnesia for the events of the acute illness. Relapses occur in up to 25% of non-paraneoplastic patients suggesting that long-term immunosuppression may be helpful.25 26
Antibody-mediated encephalomyelopathies and cerebellar ataxia
SPS and progressive encephalomyelopathy with rigidity and myoclonus (PERM)
Antibodies have been identified in a range of conditions associated with rigidity, spasms and abnormal movements. In these patients it may be relevant to request testing of antibodies to GAD, glycine receptor and the onconeural antibodies (table 3).
Antibodies to GAD
SPS is a rare disorder characterised by muscle rigidity and episodic spasms, which typically affect the lumbar, thoracic, paraspinal and proximal leg muscles. Respiratory function can be compromised leading to increased mortality in this patient group. Patients can have high levels of anxiety and panic attacks and are often initially thought to have a non-organic disorder. Antibodies to GAD are found in the non-paraneoplastic form of the disease,29 whereas amphiphysin antibodies are found in forms associated with neoplasm, most commonly of the breast or lung.2 EMG demonstrates continuous low frequency firing of motor units. Treatment includes medication directed against the muscle spasm and rigidity such as benzodiazepines and baclofen, antiepileptics and immunotherapies.30
Antibodies to GAD are also found in PERM as well as other forms of SPS such as stiff limb syndrome. In patients with PERM, rigidity and spasms are preceded or accompanied by brainstem signs (often ataxia and vertigo) and sensory disturbance, followed by generalised myoclonus and autonomic features.31 Progression of symptoms can be variable, from a protracted course of typical SPS with some features of encephalomyelitis, to a more rapid deterioration involving the brainstem and cranial nerves until death occurs. For this reason, it is important to monitor patients closely to observe for such deterioration. In PERM, CSF may show lymphocytic pleocytosis and oligoclonal bands, while imaging, EMG findings and treatment approaches are similar to those in SPS.31
Antibodies to glycine receptor
Antibodies to the glycine receptor were first identified in one case of PERM, associated with an excessive startle response. The patient responded to aggressive immunotherapy with intravenous immunoglobulin, prednisolone, plasma exchange and intravenous cyclophosphamide and returned to work part-time with only some residual spinal rigidity and bilateral ptosis.32 Several other SPS or PERM patients with antibodies to glycine receptors have now been identified.33,–,35 The clinical spectrum and treatment responses are variable; only one patient had a tumour and he made a complete recovery after thymectomy.34 One young man, who also had NMDAR-Abs, died unexpectedly.35
GAD-Ab cerebellar ataxia
Cerebellar ataxia has many causes, of which paraneoplastic cerebellar degeneration is the best known, typically associated with antibodies to the onconeural antigens Yo or Hu (and less frequently amphiphysin or Tr), and with a poor outcome.2 However, paraneoplastic cerebellar ataxia can also be associated with antibodies to the voltage-gated calcium channel. These are usually in the context of small cell lung cancer, and treatment responses are disappointing.2 36 Nevertheless, cerebellar ataxia is the second most common neurological condition associated with antibodies to GAD.23 The patients predominantly present with gait ataxia, some also have limb ataxia, dysarthria and nystagmus. Although antibodies to GAD are also noted in insulin dependent diabetes mellitus, the levels are typically much higher in patients with neurological diseases (>1000 pM but not always measured quantitatively in routine laboratories). Clinical improvement and increased cerebellar perfusion have been found in some patients with GAD-Ab and cerebellar ataxia treated with intravenous immunoglobulins.37
A recent study of 52 patients with subacute onset of ataxia identified antibodies to CASPR2 in seven, antibodies to voltage-gated calcium channels in three, and to GAD in only one (E Becker, L Zuliani, A Vincent, unpublished data). Interestingly, only one patient (who also had LE) had high VGKC-complex antibodies by the RIPA. None of the seven patients had tumours identified, including four who had been followed up for many years. This work needs to be extended but it seems that there may be several different antibodies associated with autoimmune non-paraneoplastic forms of ataxia, and this is an area that is likely to expand in the future.
Wider implications, general considerations and immunotherapies
Increasingly, each of the antibodies discussed above is beginning to be found in patients with a wider spectrum of clinical features, or in those in whom an autoimmune aetiology might not have been expected. First, there has been a steady trickle of cases of patients with cryptogenic epilepsy who have VGKC or GAD antibodies,38 39 although in many of these studies the long-duration of the condition at serum testing raises questions regarding a primary or secondary role of the antibodies. Nevertheless, we have encountered patients with an acute onset of temporal lobe epilepsy (TLE)and highly positive VGKC-complex antibodies directed against both LGI1 and CASPR2 (C Buckley, S R Irani, A Vincent, unpublished data). In a recent study of new onset epilepsy, 15% had VGKC-complex Abs, strongly suggesting that these may play a primary role in some forms of epilepsy, probably both in children40 41 and adults (T Brenner, B Lang, unpublished observations). This field has been thoroughly reviewed recently.42
Clinicians should consider antibody testing in patients presenting with acute encephalitis. In a UK study of encephalitis presenting to district hospitals,43 16 of the 44 patients (36%) without an identifiable infectious aetiology, had autoantibodies to either NMDAR (9 patients) or VGKC-complexes (7 patients), making autoimmunity the third highest association with acute encephalitis. Furthermore, in a series of patients from the Queen Square intensive care unit, around a fifth of the patients with encephalopathies had NMDAR-Abs.44
Children with enecephalitis lethargica associated with NMDAR-Ab have been described in a non-paraneoplastic setting. These patients presented with dyskinesia, seizures, agitation and somnolence and many of their features were also immunotherapy-responsive.45 Although clinically similar to typical NMDAR-Ab encephalitis, their prominent dyskinesias, lack of marked cognitive changes and lowered consciousness meant that the diagnosis could have been missed.
There may also be cases with a relatively pure psychiatric disturbance. Three of 46 patients with first episode psychosis were found to have NMDAR-Ab (n=2) or VGKC-complex Ab (n=1).46 Their response to immunotherapies was variable and more detailed studies are required.
It seems likely that these observations are just the tip of the iceberg and that further studies and time will demonstrate that autoimmunity with antibodies to cell-surface receptors will become an important part of the differential diagnosis of many more neurological and neuropsychiatric syndromes.
Serum or CSF antibodies for diagnosis and management
In our experience, absolute antibody levels are almost always higher in the serum than in the CSF. Only in very rare cases are the CSF levels positive while serum levels are undetectable, and this may be because the serum is usually diluted, at least 1:20 and frequently more, whereas CSF can be tested undiluted (because of the much lower total immunoglobulin G concentration). Much more commonly, the serum values are similar or higher despite a relative serum dilution. Nevertheless, there is often (VGKC-complex antibodies) or almost always (NMDAR, GAD, glycine receptor antibodies) intrathecal synthesis of the specific antibody which means that some of the specific plasma cells have gained access to the intrathecal compartment and are synthesising the antibody there. How early in the disease intrathecal synthesis occurs and whether CSF levels are more informative than serum levels, as some believe, is difficult to say at present, especially without a direct multicentre comparison of the assays. There is at least some evidence that plasma exchange, and probably many other treatments, directly or indirectly affect both serum and CSF levels in parallel, suggesting that serum evaluation should be sufficient. These observations, however, are confounded by different timing of paired samples, and it would be helpful to have regular paired serum and CSF samples at onset, at nadir, and after ‘successful’ treatment for careful antibody measurements in order to establish the most useful measurements for clinical management in the future. For routine diagnosis in Oxford, serum is sufficient but CSF will always be tested in parallel if provided. For other centres, it would be wise to enquire before sending the samples.
Clinical phenotyping and antibody testing
Although the specificity of the antibodies for their cell-surface target appears to be very high (ie, sera positive for binding to VGKC-complex proteins, do not in the great majority of cases bind to NMDAR or GlyRs), there are rare case reports of patients who have more than one antibody35 and even with careful clinical phenotyping, it is clear that there is considerable overlap between the syndromes associated with the different antibodies. For this reason in particular, and in order to reduce costs and delay in diagnosis, multiple antigen arrays for testing should be available in the future.
Hadavi S. Stiff person syndrome. Pract Neurol 2011; 11:272–82
Management of patients with antibody-mediated CNS disease
With such a wide spectrum of disease and range of antibodies, the management of these patients is complex. As previously explained, it is crucial to rule out an underlying malignancy, with CT or positron emission tomography scanning, if appropriate. It may take time for antibody test results to become available and this has been highlighted in questions regarding patient care, particularly in an intensive care setting. Once an infective cause of encephalitis has been ruled out (ie, PCR for herpes simplex virus) immunotherapy should be considered, irrespective of cancer diagnosis or antibody results, as patients tend to respond better and may have fewer long-term sequelae with early aggressive immunotherapy.10 25,–,27 Corticosteroids, intravenous immunoglobulin and plasma exchange are suggested but other drugs including rituximab are increasingly used.
The field of antibody-mediated CNS disease is rapidly expanding. Thorough clinical assessment and appropriate antibody testing are crucial for patient care and future research. Ongoing studies should focus on identifying antibodies to these cell-surface neuronal antigens in patients with a wide range of neurological presentations, finding new antigenic targets, demonstrating antibody pathogenicity by in vitro and in vivo studies, determining the origin of antibody synthesis, and improving clinical care through appropriate immunotherapy management and formal clinical trials.
SRI was supported by the National Institute of Health Research (NIHR), Department of Health, UK, and CB by the Medical Research Council, Great Britain. The Neuroimmunology Laboratory is supported by the NIHR Oxford Biomedical Research Centre.
Competing interests AV and the Department of Clinical Neurology in Oxford receive royalties and payments for antibody assays. AV is the inventor on patent application WO/2010/046716 entitled ‘Neurological Autoimmune Disorders’. The patent has been licensed to Euroimmun AG for the development of assays for LGI1 and other VGKC-complex antibodies. AV acts as a paid consultant for Athena Diagnostics and is employed by Oxford University and University College London. AV and SRI may receive royalties for testing of VGKC-complex antibodies.
Provenance and peer review Commissioined by Charles Warlow; this paper was reviewed by Neil Scolding.
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.