Sydenham's chorea

SC, the neurological manifestation of rheumatic fever (RF), is the prototype of chorea resulting from immune mechanisms. Chorea occurs in 26% of patients with RF although this figure varies according to time and geographical area.5 Although largely confined to areas outside North America and Western Europe, it has drawn grown interest lately. This results from the controversial possibility that a similar pathogenic mechanism may be responsible for a subset of patients with Tourette’s syndrome (TS) and related conditions—the so-called paediatric autoimmune neuropsychiatric disorders associated with Streptococcus infection (PANDAS).6 Although it is outside the scope of this paper to discuss PANDAS, the currently available evidence weighs against its existence. Despite the fall in the incidence, SC remains as the most common cause of acute chorea in children worldwide.3 In fact occasional outbreaks of RF with chorea are still occurring in the USA and Australia (see Refs. 1, 2 of online supplementary material). In a prospective study of acute movement disorders in children in a tertiary paediatric hospital in Australia, the authors identified SC in 5.7% of all cases.7

The usual age at onset of SC is 8–9 years, although there are reports of adult onset of SC.5 One important clinical finding is the observation that SC is very rarely seen in those below age 5 years.8 Typically, patients develop this disease 4–8 weeks after an episode of group A β-haemolytic streptococcus (GABHS) pharyngitis. Chorea rapidly spreads, becoming generalised but 20% of patients remain with hemichorea.5 ,8 It remains to be understood why movement disorders, probably related to humoral autoimmune mechanisms, are often unilateral. Patients with SC often display other motor findings. Motor impersistence, particularly noticeable during tongue protrusion and ocular fixation is quite common and the muscle tone is usually decreased. In severe and rare cases (8% of all patients seen at the Movement Disorders Clinic of the Federal University of Minas Gerais (MDC-UFMG)), this is so pronounced that the patient may become bedridden, a condition known as chorea paralytica, originally described by Gowers in the 19th century.9 Although there are reports of occurrence of tics in SC (see Ref. 3 of online supplementary material), they are not easy to diagnose in patients with hyperkinesias. In a cohort of 89 patients with SC followed up at the MDC-UFMG, we identified simple vocalisations in 7 patients almost always in association with facial and pharynx chorea. This suggests that tics with features similar to TS are rare in SC.10 There is evidence that many patients with active chorea have hypometric saccades, and a few of them also show oculogyric crisis.5

There has also been growing interest in the behavioural abnormalities present in SC. At the MDC-UFMG, Maia et al11 found that obsessive compulsive behaviour, obsessive compulsive disorder (OCD) and attention deficit and hyperactivity disorder were more frequent in the SC group (19%, 23.2%, 30.4%, respectively) than in the healthy patients (11%, 4%, 8%, respectively) and in the RF without chorea group (14%, 6%, 8%, respectively). A more recent systematic review confirms that all studies of behavioural changes in patients with SC have found an association with OCD.12 In a careful investigation of psychiatric comorbidities in 50 patients with SC, we found that the most frequent psychiatric disorders observed in patients with SC were: major depression (14%); generalised anxiety disorder (16%), social phobia (24%) and OCD (24%).13 Migraine is more commonly diagnosed in patients with SC than in matched controls.14 During the 19th century Gowers had already recognised that patients with SC present with a ‘disinclination to speak’. In fact, a case–control study of patients described a pattern of decreased verbal fluency that reflected reduced phonetic, but not semantic, output.15 This result is consistent with dysfunction of the dorsolateral prefrontal-basal ganglia circuit. More recently we replicated these findings in another cohort of patients with SC. Interestingly individuals with RF, although without chorea, also had a significantly worse verbal fluency.16 Studying adults with a history of SC, we have extended this finding, showed that many functions dependent on the prefrontal area are impaired in these individuals. The conclusion of this study is that SC should be included among the various causes of dysexecutive syndrome.17 Finally, it must be kept in mind that SC is a major manifestation of RF: 60–80% of patients display cardiac involvement, particularly mitral valve dysfunction, whereas the association with arthritis is less common, seen in 30% of patients. However, in ∼20% of the patients, chorea is the sole finding.5

The pathogenesis of SC is related to existence of molecular mimicry between streptococcal and central nervous system antigens. In 1956 Taranta and Stollerman established the casual relationship between infection with group A β-haemolytic streptococci and the occurrence of SC.18 Based on the assumption of molecular mimicry between streptococcal and central nervous system antigens, it has been proposed that the bacterial infection in genetically predisposed patients leads to the formation of cross-reactive antibodies that disrupt the basal ganglia function. Several studies have demonstrated the presence of such circulating antibodies in 50–90% of patients with SC.19 ,20 A specific epitope of streptococcal M proteins that cross-reacts with basal ganglia has been identified.21 In one study we found that all patients with active SC have antibasal ganglia antibodies demonstrated by ELISA and western blot.20 An investigation from another group confirmed that antibodies from patients with SC bind to neuronal surface.22 Another study showed that these autoantibodies target neuronal tubulin (see Ref. 4 of online supplementary material). These targets of the circulating antibodies are difficult to reconcile with the notion that they may interfere with neuronal function leading to chorea since binding is not restricted to basal ganglia neurons. In fact, even in the pioneer study of Husby et al19 and in our own investigation,20 the antibodies did not specifically target neurons of the circuitry related to the genesis of chorea. For this reason many authors prefer to call them antineuronal rather than antibasal ganglia antibodies. Another issue that raises concerns about the pathogenic value of these antibodies is their presence in healthy controls and individuals with other conditions such as PD and Huntington's disease (HD).19 ,20 Indeed there is little sense in using the terminology's positivity and negativity for antibasal/antineuronal antibodies. Most of these studies have employed the ELISA technique in which the difference between healthy controls and patients with SC is quantitative. This means that every single individual, either in control or SC group, has circulating antibodies. Positivity is usually defined as having a titre greater than the mean plus two SDs of the control group.20 The test of antibasal ganglia/antineuronal antibodies, is not commercially available, being just performed for research purposes. The main reasons for this are the uncertainty of their pathogenic values as well the controversy on how to define positivity.

The data reviewed in the previous paragraph demonstrate that individuals with SC have circulating serum antibodies targeting epitopes in neurons in the central nervous system. It is a contentious issue, however, if they have a pathogenic value or merely represent an epiphenomenon. The crucial issue underlying this controversy is to determine if the antibodies interfere with neuronal function. There are studies that favour this possibility. For instance, IgM of one patient with SC was reported to induce expression of calcium-dependent calmodulin in a culture of neuroblastoma cells.23 We found a linear correlation between the increase of intracellular calcium levels in PC12 cells and antibasal ganglia antibody titre in the serum from patients with SC. These results add further support to the hypothesis that the antibasal ganglia/neuronal antibodies interfere with the intracellular metabolism of calcium.24 On the other hand, there are data that fail to demonstrate that the antibodies have a pathogenic value. In one in vivo study antibodies from patients with SC infused in the basal ganglia of rodents did not induce behavioural changes. Interestingly, though, the antibodies bound to a ∼50-kDa molecule in the striatum extract.25 These results are at odds with another study of our own group that lends support to the molecular mimicry hypothesis. We found that infusion of antibodies extracted from the sera of patients with SC into the brain of rats with 6-OH-dopamine-induced unilateral lesion of the nigrostriatal system-induced circling behaviour similar to apomorphine.26 This finding suggests that the circulating antibodies act on dopamine receptors. Interestingly, more recent studies of other group of investigators support a ‘dopaminergic hypothesis’ of the pathogenesis of SC. In their first study the authors showed that rats exposed to Streptococcus antigens displayed behavioural abnormalities reminiscent of SC. Moreover, the animals had IgG that reacted with tubulin, D1 and D2 as well as 5HT-2A and 5HT-2C receptors and caused increase of the signalling of calcium/calmodulin-dependent protein kinase II (see Ref. 5 of online supplementary material). In another study this group demonstrated that the passive transfer of the antibodies replicated the findings described in rats exposed to Streptococcus antigens.27 More data add additional support to the involvement of dopaminergic transmission in the pathogenesis of SC:26 monoclonal antibodies from patients with SC bind to D2 receptors dopaminergic neurons in transgenic;28 and there is a correlation between severity of clinical features of SC and the ratio between anti-D1 and anti-D2 receptors antibodies.29 Of note, these studies have been performed by a single group of investigators. It is necessary to have the data replicated by other researchers. In summary, although it is widely believed that an autoimmune mechanism underlies SC, it remains to be fully understood.

From the neurophysiological point of view, it is widely accepted that choreas in general are characterised by increased excitability of the motor cortex.1 However, a recent study of 16 patients with SC has demonstrated that there is a decreased output of the motor cortex similarly to what is found in HD. The reasons for this paradoxical finding remain to be determined (see Refs. 6, 7 of online supplementary material).

The current diagnostic criteria of SC are a modification of the Jones criteria: chorea with acute or subacute onset and lack of clinical and laboratory evidence of alternative cause is mandatory findings. The diagnosis is further supported by the presence of additional major or minor manifestations of RF.5 ,30 The aim of the diagnostic workup in patients suspected to have SC is threefold: (1) to identify evidence of recent streptococcal infection or acute phase reaction; (2) to search for cardiac injury associated with RF; and (3) to rule out alternative causes. However, one must be aware that the first objective is much less helpful in SC than in other forms of RF due to the usually long latency between the infection and onset of the movement disorder. For instance, the classical positivity for antistreptolysin O (ASO) antibodies, helpful in the diagnosis of other forms of acute RF, is much less in SC. This is a result of the eight or more weeks of latency between the streptococcal infection and the onset of SC, when the titres of ASO have already declined. In contrast, anti-DNAase antibodies tend to circulate for a longer period of time, which makes them more helpful to identify a previous exposure to Streptococcus. Heart evaluation (ie, Doppler echocardiography) is mandatory because of the common association of SC with carditis. Serological studies for systemic lupus erythematosus (SLE) and primary antiphospholipid antibody syndrome (PAPS) must be ordered to rule out these conditions. Spinal fluid analysis is usually normal, but it may show a slightly increased lymphocyte count. In general, neuroimaging will help rule out structural causes of chorea. CT scan of the brain invariably fails to display abnormalities. Similarly, brain MRI is often normal, although there are case reports of reversible hyperintensity in the basal ganglia area. In one study, the authors showed increased signal in just 2 of 24 patients, although morphometric techniques revealed mean values for the size of the striatum and pallidum larger than controls (see Ref. 8 of online supplementary material). Positron emission tomography and single-photon emission CT imaging may be useful tools in the work up of these patients, revealing transient increases in striatal metabolism (see Refs. 9, 10 of online supplementary material). This finding contrasts with other choreic disorders, such as HD, that are associated with basal ganglia hypometabolism.

There are no controlled studies of symptomatic treatment of SC. For most people, however, the first choice is valproic acid although other anticonvulsants, such as carbamazepine, are also found to be effective and well tolerated.31 ,32 Dopamine receptor blocking agents, usually risperidone, are for patients who fail to respond to valproic acid or those rare cases with chorea paralytica. The need for caution in the use of neuroleptics in rheumatic chorea is demonstrated by a case–control study, comparing the response to these drugs in patients with SC and TS. We demonstrated that 5% of 100 patients with chorea developed parkinsonism, dystonia or both, whereas these findings were not observed among patients with tics matched for age and dosage of neuroleptics.33 There are no controlled studies of tetrabenazine or other VMAT2 inhibitors in SC. Open label studies, however, suggest that these agents may be an efficacious and safe option in some of these patients (see Ref. 11 of online supplementary material). Steroids given as intravenous pulse therapy of methylprednisolone are reserved for patients with disabling chorea refractory to antichoreic agents or in patients who developed side-effects to neuroleptics.34 ,35 There are few reports describing the usefulness of plasma exchange or intravenous immunoglobulin in SC (see Ref. 12 of online supplementary material). Owing to the efficacy of other therapeutic agents described in the previous paragraph, potential complications and high cost of the latter treatment modalities, these options are usually not recommended. Finally, the most important measure in the treatment of patients with SC is secondary prophylaxis with penicillin or, if there is allergy, sulfa drugs up to age 21 years. The recommendation is to use penicillin G benzathine intramuscular every 21 days at the dose of 1.2 million units for patients with weight >27 kg and 0.6 million units for patients with <27 kg. There is controversy regarding the duration of the prophylaxis. In areas where RF is endemic, the WHO recommends to keep this regimen until age 21 years. Outside these regions most authors opt for a 6-month duration of treatment although there is evidence lacking as to the effectiveness of this regimen.5

SC is often described as a self-limited condition, which spontaneously comes into remission after a course of 8–9 months. However, prospective studies have shown that up to 50% of patients may remain with chorea after a follow-up of 2 years. This has been called persistent SC36 (figure 1). Moreover, despite regular use of secondary prophylaxis, recurrences of SC are observed in up to 30% of patients.36 ,37 Interestingly, in many of the recurrences there is lack of association either with streptococcus infection or even antibasal ganglia antibodies.37 ,38 The underlying mechanism of these recurrences remains to be determined. It is possible, however, that the first episode of SC induces irreversible changes in the basal ganglia that remain prone to developing chorea if exposed to non-specific agents. The persistence of cognitive changes as well as mild parkinsonism in individuals with SC in remission gives support to this hypothesis.39 This seems to be the case as well as in the well characterised chorea gravidarum, recurrence of SC during pregnancy.40

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A Kaplan-Meyer curve of the time for remission of a cohort of 108 patients with Sydenham's chorea prospectively followed up at the UFMG Movement Disorders Clinic.

SLE and related conditions

SLE and PAPS are classically described as the prototypes of autoimmune choreas.41 However, several reports show that chorea is seen in no more than 1–2% of large series of patients with these conditions42 ,43 (see Ref. 13 of online supplementary material). A prospective observation of a cohort of 370 patients with SLE shows that chorea is indeed a rare finding in this condition since no patient was found to have it.44 In a series of 41 patients with SLE, the authors identified 4 patients with movement disorders (2 with parkinsonism and 1 each with myoclonus and chorea).45 Of clinical relevance is the finding that chorea may precede the onset of other findings of SLE in up to 22% of patients.46 In our own centre a systematic and careful evaluation of 54 consecutive patients with SLE we found that movement disorders were rare, 1 patient had chorea and the other one, parkinsonism. In contrast, abnormal findings of the Montreal Cognitive Assessment, Frontal Assessment Battery, semantic fluency and Yale-Brown Obsessive Compulsive Scale were found in 61%, 20%, 9% and 9% of individuals, respectively. Taking together abnormalities found on all tests, 98% of all patients met criteria for neuropsychiatric lupus.47 In summary, behavioural abnormalities are very common in SLE whereas movement disorders are uncommon and when they happen it is both chorea and myoclonus and even parkinsonism. The conclusion derived from these studies is that SLE is almost always a neuropsychiatric condition.

The pathogenesis of movement disorders and other features of neuropsychiatric lupus is poorly understood. It is possible that several mechanisms may account for these abnormalities. Several studies demonstrate the coexistence of chorea and other neuropsychiatric syndromes with antiphospholipid antibodies. For instance, Sanna et al43 found positivity for these antibodies in 185 of their 323 consecutive patients with neuropsychiatric SLE. Another study reported that 90% of patients with SLE with chorea test positive for antiphospholipid antibodies.48 Interestingly, there are individuals without criteria for SLE who have circulating antiphospholipid antibodies. Some of them display neurological (eg, stroke, chorea and others) and non-neurological features such as repeated abortions and deep venous thrombosis. These patients are labelled as having PAPS.1 ,41 It is hypothesised that antiphospholipid antibodies induce vascular changes resulting in CNS abnormalities. Indeed immunological studies show that antiphospholipid antibodies associated with chorea target β2-glycoprotein I, causing thrombotic phenomena (see Ref. 14 of online supplementary material). The usual finding of multiple small hyperintense lesions in the white matter of patients with SLE with neuropsychiatric syndromes supports the role of vascular lesions in their pathogenesis. In summary, there is evidence supporting the notion that antiphospholipid antibodies lead to vascular changes in the brain causing chorea and other movement disorders. Other lines of evidence suggest that patients with neuropsychiatric SLE have circulating antibodies, including anti-N-methyl-d-aspartate receptor (NMDAR) subunit NR2A/B, that bind to the neuronal surface leading to their dysfunction.49 ,50 The latter mechanism seems to be particularly related to cognitive dysfunction in SLE. Other potential mechanisms are disruption of the blood–brain barrier, neuroendocrine-immune imbalance, tissue and neuronal damage mediated by proinflammatory cytokines (Interleukin (IL)-1, IL-6, IL-8, IL-17 and tumour necrosis factor) as well as direct neuronal cell death (see Ref. 15 of online supplementary material).

Similarly to SC, there are no controlled studies of treatment of chorea associated with SLE or PAPS. The current recommendations primarily reflect open label studies and opinion of experts. For symptomatic control of the chorea, the mainstay is the use of neuroleptics. As there is widespread acceptance of the underlying thrombotic changes, most experts use oral anticoagulant agents. Immunosuppressive measures, such as steroids and other agents, are also commonly used.

Paraneoplastic chorea

Autoimmune chorea has also been reported in the context of paraneoplastic syndromes. In the study of adults 36 consecutive cases with autoimmune chorea, paraneoplastic chorea was identified in 14 individuals.4 As it happens with other paraneoplastic manifestations, onset of chorea usually antedates the discovery of the underlying neoplasm for periods of time that can be quite long. From a clinical point of view, suspicion of paraneoplastic nature of chorea should be raised when a patient has subacute onset of chorea in association with constitutional symptoms such as unexplained weight loss as well the association with other neurological findings. This is illustrated by a study of 1 of the largest cohorts of patients with paraneoplastic chorea, in which the authors reported that 13 patients often presented with associated findings such as dystonia, polyneuropathy, encephalitis, psychiatric disturbances or visual defects.51 This form of chorea is more commonly, but not exclusively, seen in patients with small-cell lung carcinoma or adenocarcinomas (see Refs. 4, 16, 17 of online supplementary material). The antibodies described in relationship with paraneoplastic chorea are anti-Hu (also known as ANNA-1), anti-CRMP5, anti-LGI1 and CASPR2, with the two first accounting for the majority of the cases.4 Unfortunately overall the prognosis of this type of chorea is grim both in terms of probability of remission of the movement disorder as well as survival due to the underlying disease. The exception to the latter are the cases related to LGI1 and CASPR2 antibodies, which often have a good outcome. However, similar to what happens with other conditions associated with antibodies that target intraneuronal antigens, in general paraneoplastic chorea tends to remain unabated. Regarding symptomatic management, control of the chorea can be achieved either with tetrabenazine or dopamine receptor blocking agent (see Refs. 1, 11 of online supplementary material).

Anti-NMDAR encephalitis

Encephalitis caused by antibodies against the GluN1 subunit of the NMDAR is one of the most common sporadic encephalitis worldwide. In one study it even exceeded the incidence of herpes simplex encephalitis in young patients.52 It usually occurs in young children and adults but it can affect patients of all ages. The typical clinical picture is characterised by prodromal influenza-like symptoms, followed by a psychiatric phase with psychosis, disinhibition, sleep disorders, catatonia, seizures and decrease of consciousness level. In the second phase 80% of patients develop movement disorders that are usually a mixture of stereotypies, limb and oromandibular dystonia, blepharospasm, limb and facial myorhythmia, athetosis, tremor, ataxia and chorea.53–57 In a description of phenomenology of nine children with this condition, chorea was identified in four of them but always in association with other movement disorders.58 In the final phase, there is a gradual improvement of psychiatric and neurological findings. There is, however, great latitude of the clinical features with patients displaying a variable combination of the previously mentioned features. Less than 5% of individuals may have isolated symptoms such as movement disorders, psychiatric symptoms or seizures. Recurrences are also seen in up to 12% of patients with anti-NMDAR encephalitis.53–57 There is evidence that the anti-NMDAR antibodies play a pathogenic role in the development of clinical features of the condition. A recent investigation demonstrated that they interfere with memory and behaviour in mice.59 Ovarian teratoma is associated with anti-NMDAR encephalitis in 50% of adult women and 1/3 of teenage girls. Coexistent tumour is rare in men and girls with age <14 years.53 From a movement disorders point of view, there are little, if any, differences between anti-NMDAR encephalitis patients with or without tumours. However, in the former group, the following features were more common: confusion, agitation, amnesia, spontaneous reduction in conscious level and higher levels of initial NMDAR antibodies.54 ,55 Interestingly it has been demonstrated that the rare cases of late-onset chorea in children with herpes simplex encephalitis have been associated with circulating anti-NMDAR antibodies (see Ref. 18 of online supplementary material). Just one-third of patients have abnormalities of brain MRI with non-specific cortical or subcortical hyperintense signal on fluid-attenuated inversion recovery/T2 sequences53 (see Ref. 16 of online supplementary material). Similarly to all forms of chorea and movement disorders discussed in this article, there are no controlled studies of treatment of anti-NMDAR encephalitis. There is agreement among experts, however, treatment should be started as early as possible since delays are associated with worse prognosis.55 First-line treatment comprises IV IgG, steroids or plasmapheresis whereas other options (rituximab and/or cyclophosphamide) are left for the 50% of less of individuals who fail to respond to the first options.60 Initial data suggest that the cerebrospinal fluid concentration of the chemokine CXCL13 may be useful as a biomarker of treatment response since prolonged or secondary elevation is associated with limited response to treatment and relapses (see Ref. 19 of online supplementary material).

Concluding remarks

The list of autoimmune causes of chorea continues to grow: there are recent reports of association with human papilloma virus vaccine, LGI1 antibody, anti-GABAB receptor antibody, sarcoidosis, Sjögren syndrome and Behçet's disease (see Refs. 20–25 of online supplementary material). Of note in all these conditions, the usual clinical picture does not include chorea. One good example is the LGI1 antibody, a condition previously known as antivoltage gated potassium channel. The typical clinical is the onset of the so-called faciobrachial dystonic seizures followed by limbic encephalitis. Interestingly this is usually an idiopathic condition rarely associated with underlying neoplasms. This is another disease where early immunosuppressive treatment results in better prognosis (see Refs. 26, 27 of online supplementary material).

In conclusion, although choreas of autoimmune origin are rare conditions, they are of growing interest both to researchers and practitioners. For the former they represent a window of opportunity to understand how immune mechanisms can impair the function of the basal ganglia. For the clinicians, their recognition is of practical importance since they are diseases amenable to specific treatments.