Article Text

Download PDFPDF
Are antibasal ganglia antibodies important, and clinically useful?
  1. Davide Martino1,
  2. Andrew Church2,
  3. Gavin Giovannoni3
  1. 1Clinical Research Fellow Department of Neurological and Psychiatric Sciences, University of Bari, Bari, Italy
  2. 2Senior Scientist Department of Neuroinflammation, Institute of Neurology, University College London, London, UK
  3. 3Professor of Neurology Institute of Cell and Molecular Science, Queen Mary University London, Head of Department, Department of Neurology, Barts and the London NHS Trust, London, UK
  1. Correspondence to:
 Dr G Giovannoni
 Department of Neurology, Barts and the London NHS Trust, The Royal London Hospital, Whitechapel, London E1 1BB, UK; g.giovannoni{at}

Statistics from

Request Permissions

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.

In the last decade there has been controversy over the existence and clinical significance of autoantibodies that react with the basal ganglia, referred to as antibasal ganglia antibodies (ABGAs) (some investigators have referred to them as anti-neuronal antibodies). The spectrum of disorders associated with ABGAs has recently been widened to include several neuropsychiatric syndromes; in common is their association with recent streptococcal infection, which suggests that these syndromes are due to an aberrant immune response triggered by streptococcal infection. Like other emerging diseases it will take time to establish causation, if any, and for these disorders to be accepted as real disease entities. Because they may turn out to have a considerable clinical impact, they clearly need to be better understood, as does ABGA testing in clinical practice.


ABGAs are autoantibodies that cross-react with human brain tissue, with the highest binding specificity against extracts from the caudate and putamen.1 They were originally detected using immunofluorescence microscopy (fig 1) in patients with Sydenham’s chorea,2 the prototypical post-streptococcal disease of the central nervous system (CNS). ABGAs can also be detected by Western blotting of an antigen preparation of human or rodent basal ganglia, or using enzyme-linked immunosorbent assays (ELISA).1 The sensitivity and specificity of the ELISA when using a crude antigen preparation is inferior to Western blotting.1 The performance of the ELISA may be improved in the future with the use of recombinant autoantigens.3

Figure 1

Immuofluorescence microscopy and Western blot of ABGAs. (A) Normal control diluted 1/50 tested against human basal ganglia with no specific staining (magnification ×200). (B) Sample from a patient with Sydenham’s chorea diluted 1/50 and tested against human basal ganglia tissue; IgG staining of axons (arrows) (magnification ×200). (C) Western blots showing serial dilution of strongly positive ABGAs in patients with post-streptococcal movement disorder.


ABGAs are associated with a wide spectrum of post-streptococcal neuropsychiatric disorders (table). Most of these disorders are considered to be related to dysfunction in the cortico–striato–thalamo–cortical circuitry of the brain which controls motor, emotional and cognitive domains. This causes a complex array of neurological (mainly movement disorders) and psychiatric symptoms (mainly, behavioural disorders with features such as obsessive-compulsive behaviour and conduct problems, or anxiety/depression states).4 Two post-streptococcal disease entities are well recognised, namely Sydenham’s chorea (now universally accepted as occurring in association with rheumatic fever) and, more recently, post-streptococcal acute disseminated encephalomyelitis (ADEM) which includes a variant manifesting with an encephalitis lethargica-like syndrome.5 Post-streptococcal ADEM presents with a movement disorder and prominent psychiatric features, and on brain MRI the patients have inflammatory basal ganglia lesions (fig 2). The encephalitis lethargica-like variant (fig 3) manifests with parkinsonism, sleep disturbances (particularly hypersomnolence), hyperkinesias and psychiatric symptoms, and may be associated with focal basal ganglia lesions on MRI. However, the other conditions on the list of putative post-streptococcal neuropsychiatric disorders are not well accepted: post-streptococcal childhood onset obsessive-compulsive and/or tic disorders, generally known as the paediatric autoimmune neuropsychiatric disorders associated with streptococcal infections (PANDAS) phenotype, Tourette’s syndrome,6 obsessive-compulsive disorder7 and adult onset tics and dystonia.8 PANDAS has generated the most controversy and remains the subject of an ongoing, and at times acrimonious, debate.9,10

Figure 2

Brain MRI of a case of post-streptococcal acute demyelinating encephalomyelitis. A young boy presented 10 days after an upper respiratory tract infection with encephalopathy, dystonia and pyramidal signs in all four limbs. He was intubated and ventilated for four days. Intravenous methylprednisolone followed by oral prednisolone resulted in clinical improvement and recovery. Streptococcal serology was positive, with normalisation on recovery. MRI of the brain showed extensive inflammatory lesions in caudate and putamen bilaterally, shown as hyperintense signal lesions on T2-weighted images (arrows).

Figure 3

The encephalitis lethargica syndrome is a neuropsychiatric syndrome with a movement disorder—either hypokinetic or hyperkinetic—prominent psychiatric features and a sleep disorder. This constellation of symptoms and signs results from pathological involvement of the diencephalon and mid-brain structures.

Common to all these disorders is basal ganglia dysfunction.4 Although well-defined disorders of the basal ganglia such as Huntington’s and Wilson’s diseases are associated with “classic phenotypes”, both can have atypical presentations that cover the full phenotypic spectrum associated with basal ganglia dysfunction. It is therefore unlikely that basal ganglia dysfunction occurring as a result of autoimmunity should be any different; in our opinion the wide clinical phenotype associated with post-streptococcal neuropsychiatric diseases is to be expected.

In Sydenham’s chorea, pathological abnormalities are localised to the basal ganglia, and include cellular infiltration and neuronal loss, with relative sparing of other brain areas.4 As a result of these changes an encephalitic pathogenesis has been proposed; however, postmortem data from cases of acute Sydenham’s chorea are rare and are probably biased towards unusually severe forms.4 The pathological features of post-streptococcal neuropsychiatric diseases and its encephalitis lethargica-like variant are not dissimilar to Sydenham’s chorea, with perivascular lymphocyte and plasma cell cuffing predominantly involving the mid-brain and basal ganglia.5 There are no consistent data on the neuropathology of the other syndromes associated with ABGAs, including PANDAS, Tourette’s and obsessive-compulsive disorder.

Structural brain imaging is normal in most cases of Sydenham’s chorea and PANDAS, casting doubt on whether widespread neuronal loss is an important feature of these diseases. But this does not rule out subtle changes. The focal inflammatory lesions in Sydenham’s chorea are predominantly localised to the basal ganglia on MRI and, in most cases, are partially reversible with clinical remission.4 Volumetric imaging shows enlarged caudate nuclei during acute Sydenham’s chorea compared to control subjects, consistent with focal inflammation.11 Similarly, some functional imaging studies have demonstrated basal ganglia hyperperfusion and hypermetabolism during the acute phase of Sydenham’s chorea, although this has not been observed in all studies.4 In post-streptococcal ADEM 80% of patients show hyperintense basal ganglia lesions compared to only 18% of cases of ADEM not associated with streptococcal infection. This fits well with the observation that the post-streptococcal form of ADEM is phenotypically different from the other forms of ADEM; there is a higher frequency of movement disorders, particularly dystonia, and in about 70% of these patients there is a behavioural syndrome with emotional lability, inappropriate laughter, separation anxiety, confusion and hypersomnolence.12 In contemporary cases of encephalitis lethargica, conventional MRI shows inflammatory changes localised to the deep grey matter in 40% of patients.5 When acute, these lesions often show diffuse enhancement, which may be associated with reduced dopaminergic innervation of the basal ganglia (fig 4).

Figure 4

MRI and DAT scan. A middle-aged woman with encephalitis lethargica presented with double vision, headache and a behavioural disorder, followed by increasing confusion and reduced level of consciousness due to hypersomnolence. She later developed oculogyric crises and hiccoughing. On recovery she was parkinsonian, with rigidity and bradykinesia, with superimposed tics and dystonic posturing of the right arm. Unfortunately, she was left with chronic obsessive-compulsive behaviour, anxiety, panic attacks and dysthymia. The MRI study during the acute encephalitic crisis showed bilateral swelling of the striatum, with associated signal change on the T2- (left) and proton density-weighted images (centre). These areas were shown to enhance diffusely after the administration of gadolinium (right). The abnormal signal change also extended into the posterior hypothalamus and mid-brain (images not shown). On the far right is an abnormal striatal dopamine transporter (DAT) study, with [123I]beta-CIT ([123I]2beta-carbomethoxy-3beta-(4-iodophenyl)tropane) single photon emission computed tomography (SPECT) showing bilaterally reduced and asymmetrical dopamine transporter density in the striatum. This patient’s serum was positive for ABGAs.


Post-streptococcal neuropsychiatric diseases are believed to be auto-aggressive immune-mediated disorders, triggered by exposure to streptococci. A popular, albeit unproven, hypothesis to explain their pathogenesis is the “molecular mimicry model” (fig 5). Although cell-mediated immunity and superantigen-mediated processes are involved in other post-streptococcal inflammatory disorders, most attention to date has focused on antibody-mediated effects in post-streptococcal neuropsychiatric diseases. It is conceivable that pathogenic autoantibodies produced in the periphery enter the CNS through an intact blood-brain barrier, or alternatively an autoimmune response occurs within the CNS after the relevant antigen-specific cells enter the CNS. Approximately 40% of patients with Sydenham’s chorea show intrathecal synthesis of oligoclonal IgG.13

Figure 5

The current hypothesis to explain the pathogenesis of post-streptococcal neuropsychiatric disorders is the “molecular mimicry model” in which people with a genetic susceptibility develop a cross-reactive or autoimmune response to self-antigens after an appropriate immune response to inciting bacterial antigens. These bacterial antigens are homologous with human antigens, hence the term “molecular mimicry”.

ABGAs recognise several protein bands (fig 1); the four main ones are 40, 45 and 60 and 98 kDa.14 The 40 kDa antigen is aldolase C (neuron-specific), the 45 antigen is gamma-enolase, and the 60 kDa antigen is pyruvate kinase.3 We also speculate that a fourth band against a 98 kDa antigen is a dimer of gamma and alpha-enolase. Gamma enolase-reactive ABGAs crossreact with alpha-enolase.3 Interestingly, alpha-enolase has been implicated as an autoantigen in rheumatic fever and other autoimmune diseases.15 Although these cytosolic antigens are glycolytic enzymes involved in energy homeostasis, they are also located on the neuronal surface3,15 where they appear to have “moonlighting” or alternative functions. For example, enolase on the surface of neurons acts as a receptor for plasmin/plasminogen15 which, interestingly, provides trophic support to mesencephalic dopaminergic neurons.16 Membrane neuronal aldolase provides local membrane ATP production, and is thought to monitor oxidative stress.3 Pyruvate kinase acts as thyroid hormone (T3) binding protein; binding of T3 to pyruvate kinase inhibits enzymatic activity, suggesting that this process may be involved in the control of some cellular metabolic effects induced by thyroid hormones.17 In summary, membrane glycolytic enzymes are closely involved in energy provision and maintenance of ion channels, trophic support and other functions. Disrupting their activity via antibody-mediated mechanisms might lead to neuronal dysfunction. All three of the major candidate autoantigens have protein homologues in streptococci expressed on their surface and show structural homology with the human isoforms of up to 49%, which makes them good potential “molecular mimics”.3

Evidence for an alternative autoantigen has been presented by Kirvan and colleagues, who tested monoclonal antibodies from human hybridomas derived from a patient with Sydenham’s chorea.18 One was highly specific to the mammalian lysoganglioside GM1 and to N-acetyl-beta-D-glucosamine (Glc-NAc), the dominant epitope of the group A beta haemolytic streptococci surface carbohydrate. Both the monoclonal antibody and the patient’s sera induced calcium/calmodulin-dependent protein kinase II (CAMK-II) activity, which is involved in neuronal intracellular signalling and in neurotransmitter release.18 Antibodies to Glc-NAc from PANDAS patients’ sera showed the same CAMK-II modulation in 75% of cases, although the crossreactivity with lysoganglioside GM1 was also observed in 23% of children with non-PANDAS tics or obsessive-compulsive disorder.19 Unfortunately, no healthy control group was included in this study which makes it difficult to assess the diagnostic utility of anti-lysoganglioside GM1 antibodies.19

To establish an antibody-mediated autoimmune disease, specific criteria need to be fulfilled:20

  • First, autoantibodies should be present in the serum or cerebrospinal fluid of most patients with the condition and should not be found in healthy or other appropriate control subjects. With the widening clinical spectrum of disorders associated with ABGAs their sensitivity and specificity need to be assessed in much larger groups of patients. Population-based studies of children presenting with obsessive-compulsive disorder and/or tic disorders are required. Similarly, studies are also required to assess the specificity of anti-Glc-NAc/GM1 antibodies in Sydenham’s chorea, PANDAS and related disorders.

  • Second, autoantibodies should bind to the target antigen(s) at the key site of pathology: anti-Glc-NAc/GM1 antibodies from Sydenham’s chorea patients bind human caudate and putamen, and pre-adsorption with lysoganglioside GM1 inhibits this binding. Whether this is specific for Sydenham’s chorea needs to be established as there are no data from healthy sera. Evidence to verify the cellular and subcellular targets of ABGAs in the human striatum is also required.

  • Third, the injection of serum or purified immunoglobulin should transfer the disease phenotype to experimental animals (“passive transfer”). In fact, two studies showed that infusing serum immunoglobulin from patients with PANDAS into rat striatum increased stereotypical movements compared to the infusion of control antibodies.21 However, more recently a collaborative study failed to reproduce these results across three different laboratories.21 These preliminary studies all shared the same methodological limitation in that they did not select sera according to a specific antibody positivity, hence their results could be related to a pool of other non-pathogenic antibodies.22

  • Fourth, an experimental disease homologous to the human disorder should be inducible by sensitising susceptible animals with the corresponding target antigen. This has not been attempted specifically with human neuronal glycolytic enzymes or with Glc-NAc/lysoganglioside GM1. In a recent study, mice immunised with a crude group A beta-haemolytic streptococcal homogenate developed abnormal behaviours, and their serum reacted with several brain regions.23

  • Finally, plasma exchange, which removes immunoglobulins from the circulation, should have a therapeutic effect. A controlled trial of either plasma exchange or intravenous immunoglobulin (IVIg) in children with PANDAS demonstrated a significant improvement in motor and psychiatric symptoms for both therapies compared to placebo.24

All these observations suggest that ABGAs or other humoral factors may be pathogenic but more evidence is clearly required.

Whether ABGAs are pathogenic or not, their presence indicates an ongoing autoimmune process. If ABGAs are not pathogenic, they could simply be an epiphenomenon of an immunological reaction provoked by other factors (for example, other antibodies, cell-mediated cytotoxicity or a super-antigen response) or non-specific antibody synthesis secondary to neuronal damage (epitope spreading). A few studies have shown that cytokines and chemokines, mediators of inflammation, are raised in patients with these disorders,13 and it has been shown that ABGA-positive patients with Tourette’s syndrome have raised levels of soluble vascular cell adhesion molecule-1 (VCAM-1), which modulates lymphocyte migration.6


The Biomarkers Definitions Working Group defines a biological marker as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacological responses to a therapeutic intervention”.25 Biomarkers are supposed to be accurate indicators of a concurrent pathological process or, in the case of diagnostic biomarkers, a specific medical diagnosis.

In the correct clinical setting the detection of ABGAs supports the clinical diagnosis of Sydenham’s chorea and post-streptococcal ADEM. While work is in progress to set up a sufficiently reliable quantitative assay (for example, ELISA), ABGAs are currently detected by a qualitative assay (Western blotting) which simply provides a binary answer (that is, presence/absence of the specific antibody bands). With this method, Church and colleagues detected ABGAs in all patients with acute Sydenham’s chorea and in 69% of patients with convalescent Sydenham’s chorea, compared to only 12% of cases with rheumatic fever without Sydenham’s chorea and 4% of a healthy paediatric control population.1 Similarly, the detection of ABGAs in the serum and/or cerebrospinal fluid, in conjunction with microbiological and serological techniques for detecting streptococcal infections, could discriminate between post-streptococcal ADEM and other forms of ADEM; in this context, the clinical presentation and the presence of any basal ganglia lesions on MRI should make one at least suspicious of the diagnosis of post-streptococcal ADEM or encephalitis lethargica. In the setting of these two conditions the diagnostic sensitivity of ABGAs is reported as being in the order of 95%.5

In a small study Church and colleagues described 40 children presenting with a neuropsychiatric disorder following infection with group A beta haemolytic streptococci and compared their ABGA status to several control groups (healthy children, children with uncomplicated streptococcal infections, with other neurological conditions and with systemic autoimmune diseases).14 ABGAs discriminated between the post-streptococcal children and all other groups, with a sensitivity of 93%, a specificity of 97%, and in this context had positive and negative predictive values of 97% and 91% respectively.14 In separate studies of 100 cases of Tourette’s syndrome and 50 paediatric cases of obsessive-compulsive disorder, ABGAs were detected in 23% and 42% respectively, significantly more frequently than in appropriate control populations.4,7 Similar observations have been replicated by other groups (see Martino and Giovannoni for a detailed review of the literature4). However, the issue remains controversial, mainly due to the observations of the Johns Hopkins group who have repeatedly failed to detect ABGAs in patients with Sydenham’s chorea, PANDAS and Tourette’s syndrome.26 Future formal crossover testing of clinical samples from the various laboratories could help solve the controversy, as this discrepancy may be the result of important methodological differences in Western blotting techniques.27

Streptoccal infection is usually confirmed by culture or by demonstrating a raised anti-streptolysin O (ASOT) or anti-DNAse B antibody titre. One practical difficulty in diagnosing post-streptococcal neuropsychiatric syndromes is the time lag between acute infection and the onset of CNS symptoms. For Sydenham’s chorea, this interval may be as long as 2–6 months, by which time the organism is no longer detectable in the patient’s throat and the serological markers may have declined in titre or even normalised. In fact, only 60–70% of patients with rheumatic fever have serological evidence of recent streptococcal infection when they develop Sydenham’s chorea. At this stage, Sydenham’s chorea is diagnosed either by association with rheumatic fever, a history of infection compatible with streptococcal infection, and/or by excluding other diagnoses. In these latter situations the presence of ABGAs may help diagnostically. In the reported cases of post-streptococcal ADEM the latency between infection and the illness was generally days or weeks, and culture and/or serology was critical in establishing the link between streptococcal infection, the neurological syndrome, and the presence of ABGAs.5,12

Children experiencing an acute onset of a movement disorder (particularly chorea and tics) and/or a behavioural abnormality, who have detectable ABGAs in their serum, have a greater than 90% chance of having serological evidence of recent streptococcal infection.14 In cross-sectional studies of unselected cohorts of patients with Tourette’s or obsessive-compulsive disorder, raised ASOTs were more frequent in ABGA-positive than in ABGA-negative patients.4,7 This evidence favours the view that ABGAs may be an indirect marker of previous streptococcal exposure, or the possibility of exposure to related organisms, despite not being able to date such exposure.

ABGA-associated syndromes tend to be more common in children. However, ABGAs are also found in adults presenting with atypical movement disorders.8 Of 65 such patients, 42 were ABGA-positive. ABGA-positive cases were younger and tended to present with an isolated fixed limb dystonia. Other syndromes identified included tics, parkinsonism, myoclonus, chorea and ataxia. These adult-onset ABGA-positive movement disorders were more frequently associated with precipitating factors, particularly recent infections, than cases which were ABGA-negative. There was no clear relationship between these patients and streptococcal infection, although prior subclinical exposure could not be excluded. Whether or not finding ABGAs in adult patients (particularly, those with “fixed dystonia”) indicates that ABGAs lack specificity, requires further study. Classic movement disorders, however, such as primary dystonia or Parkinson’s disease, are not associated with ABGAs.8


There are well established symptomatic and prophylactic treatment strategies for Sydenham’s chorea. Antibiotic prophylaxis in subjects who have had rheumatic fever and/or Sydenham’s chorea is mandatory and is standard clinical practice. In most parts of the world group A beta haemolytic streptococci remain sensitive to penicillin which is therefore the antibiotic of choice. In subjects who cannot tolerate penicillin, macrolides are recommended despite a high risk that antibiotic resistance may develop. A recent observation suggested that antibiotic prophylaxis may be effective in reducing symptomatic exacerbations in children with PANDAS,28 but this study did not have a control group and has been heavily criticised.29 It would therefore be premature to recommend antibiotic prophylaxis in patients with PANDAS or Tourette’s who have ABGAs.

There have been no well controlled studies of intravenous immunoglobulin (IVIg) or plasma exchange in Sydenham’s chorea. In a small study of five patients treated with plasma exchange and four with IVIg,30 patients in both treatment arms improved, although the plasma exchange group improved more rapidly. Three of the four children treated with IVIg relapsed within four months of completing treatment.

Several small studies have examined the effectiveness of corticosteroids in Sydenham’s chorea. A retrospective study of eight subjects showed rapid improvement with corticosteroids.31 In another study five subjects, refractory to standard symptomatic therapy (valproate and neuroleptics), were treated successfully with intravenous methylprednisolone and then oral prednisolone.32

The only placebo controlled trial of immunomodulation (plasma exchange and intravenous immunoglobulin) in PANDAS demonstrated improvements in the patients treated with active agents compared to patients treated with sham (saline) infusions. Surprisingly, the improvement was maintained for at least a year.24 These findings were not reproduced in obsessive-compulsive patients who did not have PANDAS, suggesting that the benefit of immune modulation is restricted to the PANDAS subgroup of neuropsychiatric disorders.

Currently, we recommend that immune treatments should not be given routinely to patients with Sydenham’s chorea or PANDAS until further controlled trials establish their efficacy beyond doubt. There is however a scientific rationale for a therapeutic trial of immunomodulatory therapy in ABGA-positive patients who are significantly disabled by their disorder. Our anecdotal experience is that the sooner patients are treated after the onset of their symptoms the more likely they are to respond to this therapy.

Practice points

  • There is a group of putative autoimmune diseases of the central nervous system characterised by ABGAs associated with recent streptococcal infection.

  • ABGAs are a potentially useful diagnostic marker of immune-mediated basal ganglia disorders triggered by a streptococcal infection.

  • Patients typically present with a neuropsychiatric syndrome with both hyper- and hypokinetic movement disorders; Sydenham’s chorea is the prototypical disorder associated with ABGAs.

  • Apart from Sydenham’s chorea, the association of ABGAs with other clinical phenotypes is not widely accepted.

  • Long-term antibiotic prophylaxis is indicated in patients with Sydenham’s chorea but there is no evidence supporting this policy in other ABGA-associated disorders.

  • Immunotherapy should not be given routinely to patients with ABGA-associated disorders until further controlled trials establish its efficacy.

  • There is a scientific rationale for a therapeutic trial of immunomodulatory therapy in patients who are significantly disabled from their disorder.

  • Our anecdotal experience is that the sooner the patients are treated after the onset of their symptoms, the more likely they are to respond to immunotherapy.

Spectrum of disorders associated with antibasal ganglia antibodies


We wish to thank the Tourette Syndrome Associations of the United States and the United Kingdom, the Sophie Cameron Trust, Action Research, University of London and University College London Hospitals NHS Trust for financial support. Dr Davide Martino was supported by a European Union Marie Curie Training Fellowship. We also thank Ms Janet Alsop for her excellent secretarial support and Dr Russell Dale for supplying clinical data and MRI images for the case presented in figure 2.