The Clinical Syndrome of Paroxysmal Exercise‐Induced Dystonia: Diagnostic Outcomes and an Algorithm

Abstract

Paroxysmal exercise‐induced dyskinesia (PED) is characterized by recurrent episodes of involuntary movement disorders usually precipitated by sustained walking or running. Recently, mutations in the gene encoding for glucose transporter type 1 (GLUT‐1) were described in a number of families with autosomal dominant PED. However, the underlying etiology of PED is quite heterogeneous. We describe a large series of patients presenting with PED. Of 16 patients, we reached a conclusive diagnosis for 11 (4 patients with GLUT‐1 mutations, 4 patients with early Parkinson's disease, 2 with dopa‐responsive dystonia, and one with a psychogenic/functional movement disorder). For the remaining 5 patients, the final diagnosis remained descriptive. Although certain clinical features might allow etiological distinction between cases, clinical examination alone is not always conclusive. Based on our series, we propose a diagnostic algorithm to aid the differential diagnosis of PED.

Paroxysmal dyskinesias are a heterogeneous group of rare conditions characterized by recurrent episodes of sudden involuntary movement disorders. They are classified according to the precipitating factors into three subtypes, namely, paroxysmal kinesigenic (PKD), nonkinesigenic (PNKD), and exercise‐induced (PED) dyskinesia.1 All three forms can present with dystonic, choreic, or ballistic movements, or a mixture of those, and can involve one or more limbs, the trunk, and/or the face.1 As far as PED is concerned, lower‐limb dystonia precipitated by sustained walking or running is the most common manifestation.1, 2

Recently, heterozygous missense mutations in SLC2A1, encoding for glucose transporter type 1 (GLUT‐1), were described in four families with autosomal dominant PED.2 However, most of the PED cases described thus far appear to be sporadic and not all carry a GLUT‐1 mutation. There have been a few reports describing PED as the presenting feature in patients with early‐onset Parkinson's disease (PD).3, 4, 5 In addition, dopa‐responsive dystonia (DRD) resulting from mutations in the GTP cyclohydrolase 1 gene (GCH1) can also present as PED.6 Finally, PED has also been reported to be symptomatic.7, 8 Thus, the underlying etiology of PED is quite heterogeneous. Prompt recognition of the underlying etiology is crucial because most of these disorders are treatable, and misdiagnosis can possibly lead to inappropriate management of these patients. We aimed to address this issue in more detail, describing the clinical features and diagnostic outcomes in a large series of consecutive patients with PED, and providing a diagnostic algorithm to aid the differential diagnosis.

Patients and Methods

We searched our database with the terms “paroxysmal exercise induced dyskinesia” and “paroxysmal exercise induced dystonia” for the last 4 years (2009–2013). Twenty patients were identified, of whom 4 were excluded (2 because exercise was not reported to be the only major trigger for the attacks and 2 for insufficient data). All these patients had PED as the main and, in the majority, the only complaint for which they have been referred to a movement disorder service. We collected details on age, disease onset and duration, previous treatments, clinical examination, and further imaging, cerebrospinal fluid (CSF), and genetic investigations. All patients had provided written informed consent to publish their clinical data, also with regard of video recordings.

Results

Eight patients were male and 8 were female. Mean age at onset was 15.8 ± 12.4 years, and mean disease duration was 19.0 ± 15.6 years. Demographic and clinical features of these patients are summarized in Table 1. After extensive diagnostic workup, including MRI of the brain, dopamine transporter (DaT‐Scan) imaging, CSF investigations for glucose, lactate, pterins, and dopamine metabolites, and further genetic testing where appropriate (Supporting Table 1), we were able to reach a conclusive diagnosis for 11 patients (4 with GLUT‐1 mutations, 4 with early PD, including 1 carrying a mutation in the Parkin gene, 2 with DRD resulting from mutations in the GCH1 gene, and 1 with a psychogenic/functional movement disorder). For the remaining 5 patients, the final diagnosis remained descriptive.

Table 1.

Demographic and clinical features in our series

Case	Gender	Age	Age at Onset	Disease Duration (Years)	Family History	Other Complaints at Disease Onset	Other Signs on Examination	Body Site Involved in PED	Phenomenology of PED	Previous Treatments	Diagnosis
1	F	19	15	4	No	No	Postural jerky tremor of both hands	Left foot	Dystonia	None	DRD
2	M	24	22	2	No	Occasional migraine	No	Right foot	Dystonia	l‐dopa (+)	Unknown
3	F	23	19	4	No	No	No	Left foot	Dystonia	l‐dopa, clonazepam, AEs (−)	Unknown
4	M	42	7	35	No	No	Mild dysmetria on finger‐nose test	Right leg and foot	Dystonia and chorea	Acetazolamide, carbamazepine (−)	GLUT‐1
5	F	25	5	20	No	Occasional migraine	Difficulties in tandem gait	Both legs and feet	Dystonia	None	GLUT‐1
6	F	25	2 (epilepsy) 7 (PED)	23	No	Epilepsy	Gaze impersistence, saccadic overshoot	Left foot	Dystonia and chorea	None	GLUT‐1
7	M	47	42	5	No	No	Tics	Left foot	Dystonia	None	PD
8	M	57	47	10	No	No	No	Left foot	Dystonia	Carbamazepine (−)	Unknown
9	M	51	16	35	No	No	Brisk reflexes of lower limbs	Right foot	Dystonia	Carbamazepine (−)	Unknown
10	F	51	20	31	No	Restless leg syndrome	No	Right foot	Dystonia	None	PD (PARK2)
11	F	25	20	5	No	No	No	Both feet	Dystonia	None	Psychogenic
12	M	34	5	29	No	Epilepsy	Cerebellar, pyramidal, neuropathy	Both feet (more severe attacks could spread to the arms)	Dystonia	AEs (+ on seizures, but not on PED)	Unknown
13	M	38	36	2	No	Slowness	Brisk reflexes lower limbs, bradykinesia	Right foot	Dystonia	None	PD
14	F	43	41	2	No	No	Mild dystonia of toes in the left foot	Left foot	Dystonia	Botulinum toxin (++)	PD
15	M	22	7	15	No	Right‐hand clenching	Mild dystonic posturing of the left foot	Left Foot	Dystonia	l‐dopa (+++)	DRD
16	F	19	1 (hypothonia, epilepsy) 17 (PED)	18	No	Hypothonia at birth, epilepsy	Mild mental retardation, spasticity lower limbs, mild ataxia	Left leg> right leg	Dystonia and chorea	AEs (+ on seizures, not on PED)	GLUT‐1

PED, paroxysmal exercise‐induced dyskinesia; F, female; M, male; DRD, dopa‐responsive dystonia; PD, Parkinson's disease; PARK2, PD resulting from mutation in the Parkin gene; AEs, antiepileptic drugs; +, mild benefit; ++, moderate benefit; +++, marked benefit; −, no effect.

After exclusion of the patient with a psychogenic/functional movement disorder, further comparisons have been made. Table 2 summarizes clinical and investigations results in our patients, stratified according to their diagnosis. Patients with GLUT‐1 had a younger age at onset and were more frequently found to have cerebellar findings on examination than other groups. Moreover, these patients were more likely to feature both dystonia and chorea during their attacks, whereas DRD, PD, or idiopathic PED patients showed, almost invariably, a pure dystonic phenotype. On the other hand, GLUT‐1 patients did not have interictal extrapyramidal features on examination (i.e., tremor, bradykinesia, or dystonia), which could be instead disclosed in patients with DRD or PD.

Table 2.

Clinical comparisons among groups stratified according their diagnosis

	Patients With GLUT‐1 (n = 4)	Patients With DRD (n = 2)	Patients With PD (n = 4)	Patients With Idiopathic PED (n = 5)
Age at onset (mean ± standard deviation)	3.75 ± 2.75	11 ± 5.65	34.7 ± 10.1	21.8 ± 15.4
Sex (M:F)	1:3	1:1	2:2	4:1
Epilepsy (no. of cases, %)	2, 50	0, 0	0, 0	1, 20
Other complaints at onset (no. of cases, %)	2, 50	1, 50	2, 50	1, 20
Cerebellar findings on examination (no. of cases, %)	4, 100	0, 0	0, 0	1, 20
Pyramidal findings on examination (no. of cases, %)	1, 25	0, 0	1, 20	2, 40
Other findings (tremor/dystonia/bradykinesia) on examination (no. of cases, %)	0, 0	2, 100	2, 50	0, 0
Body site involved in PED
Foot (no. of cases, %)	1, 25	2, 100	4, 100	4, 80
Foot and Leg (no. of cases, %)	3, 75	0, 0	0, 0	1, 20
Phenomenolody of PED
Dystonia (no. of cases, %)	1, 25	2, 100	4, 100	5, 100
Dystonia and chorea (no. of cases, %)	3, 75	0, 0	0, 0	0, 0

DRD, dopa‐responsive dystonia; PD, Parkinson's disease; PED, paroxysmal exercise‐induced dyskinesia; M, male; F, female.

Discussion

We present here the diagnostic outcome of one the largest cohorts of patients with PED, supporting previous evidence that the underlying etiologies are heterogeneous.2, 3, 4, 5, 6, 7, 8, 9

Based on the presented data, we would argue that certain clinical features might allow for etiological distinction between cases. Patients carrying a GLUT‐1 mutation were much younger than patients in other groups (virtually all GLUT‐1 cases develop the disease during childhood) and were more likely to show cerebellar involvement on examination, confirming that GLUT‐1 can affect multiple systems in the brain.2 Conversely, interictal extrapyramidal signs were never observed in GLUT‐1 patients (whereas they could be observed in some patients with DRD or PD). In a number of patients, there was a concomitant history for epilepsy. Moreover, as to the phenomenology of PED, they were more likely to feature a dyskinetic form of PED, with both dystonia and chorea involving one or both legs, whereas pure dystonic PED involving one foot were more likely observed in DRD or PD patients (see Video). However, our sample size is limited and therefore definitive conclusions cannot be drawn. With a view to age at onset, for instance, we acknowledge that DRD patients can develop symptoms very early in childhood as also patients with GLUT‐1. On the other hand, onset in adulthood would render the diagnosis of GLUT‐1 unlikely. Therefore, age at onset can be useful, but not always conclusive, information to aid in the differential diagnosis. In addition, the clinical examination alone may not always provide additional information. First, there are PED patients who show overlapping phenotypic charactistics (e.g., both patients with early‐onset PD and DRD can manifest brisk reflexes in the lower limbs). Second, patients may present with pure PED without any other subtle symptoms or signs to point to a particular etiology. For example, some of our PD patients presented only with PED without any tremor or bradykinesia. This has also been the case for the published symptomatic PED cases.7, 8 Therefore, further investigations are necessary.

Based both on the present series and the previously published cases, we would suggest a practical algorithm for the diagnosis and subsequent management of patients with PED (Fig. 1). First, secondary causes need to be excluded and a brain MRI should be performed to rule out any lesion in the basal ganglia. We advocate that the next step should be CSF investigations for glucose, lactate, pterins, and dopamine metabolites, if onset is in childhood. A lumbar puncture is an invasive, but nevertheless straightforward and low‐cost, procedure, which allows for excluding different disorders in one go. GLUT‐1 patients usually show a low glucose CSF/serum ratio (≤25th percentile), but this ratio can be within the normal range for almost 10% of these patients.10 It has been reported that raw CSF glucose levels are instead below the 10th percentile for all patients with GLUT‐1, including those with the mildest phenotypes, suggesting that this index can be more sensitive than glucose CSF/serum ratio, if age‐specific reference values are applied.10 However, CSF lactate should be concomitantly evaluated. In fact, if CSF lactate level is elevated (i.e., >90th percentile), a diagnosis of GLUT‐1 deficiency syndrome is very unlikely, and alternative diagnoses should be considered.10 Mutations in the GCH1 gene are instead invariably associated with a specific CSF pattern, featuring low levels of tetrahydrobiopterin, homovanalinaic acid, and 5‐hydroxyindolacetic acid.11 No other inherited disorder of the dopamine or serotonin pathways has been reported to manifest with PED. PD patients can also present some abnormalities in CSF dopamine and serotonin metabolites, even if this has not clearly established.12 However, PD patients do not have a reduction in pterins levels (i.e., tetrahydrobiopterin). Therefore, in the presence of normal CSF glucose and tetrahydrobiopterin, GLUT‐1 and DRD are very unlikely, respectively, and in this situation a DaT‐Scan should be performed to exclude PD. If onset is in adulthood, one may of course consider to request a DaT‐Scan before CSF investigations and to further perform the latter only in the case of normal dopaminergic imaging.

Figure 1.

Diagnostic algorithm for PED (*after 4–6 hours of fasting; blood sample should be obtained before lumbar puncture to avoid stress‐induced hyperglycemia).

At this stage, genetic testing can be tailored based on the results of the aforementioned investigations and should be pursued to reach a definitive diagnosis. This is relevant not only with regard to management, but also to genetic counseling. With a view to PD, here we show that PED can be the presenting symptom in a Parkin patient, but also in 3 other patients who tested negative for both Parkin and LRKK2 genes, suggesting that other either genetic or sporadic forms of PD are responsible of this phenotype.

After an extensive diagnostic workup, we were not able to identify any etiological diagnosis for 5 patients, in line with previous works including idiopathic cases who tested negative for GLUT‐1. Of note, there was heterogeneity among our patients with unknown causes for their PED, with 1 patient showing a complex syndromic disorder, featuring PED, cerebellar and pyramidal involvement, epilepsy, and neuropathy. This phenotype partly overlaps with the “rolandic epilepsy, paroxysmal exercise‐induced dyskinesia and writer's cramp” syndrome, a condition described in the pre‐PRRT2 gene era.13 It is extremely likely that such a syndrome (formerly linked to the locus 16p12‐11.2) was indeed the result of PRRT2 mutations.14 We have not excluded mutation in the PRRT2 gene in our patient, but we think it would be unwise to consider such a possibility. In fact, the interictal cerebellar and pyramidal features observed in our patient are against the classical phenotype observed in patients with PRRT2 mutations.14 Moreover, PED attacks, in the total absence of paroxysmal kinesigesic dyskinesia attacks, are rather uncommon in patients with PRRT2 mutations.14 Our results would suggest that different disorders might encompass PED in their phenotype. Detailed clinical clustering of such patients into homogenous groups offers the only framework for subsequent use of whole exome sequencing. This might settle the question of what these patients with “undetermined” PED have.

In summary, here we show that PED can be associated with a number of different disorders, highlighting the need for a rational diagnostic workup. Prompt recognition of the underlying etiology is crucial for subsequent management, prognosis, and genetic counseling of these patients. The algorithm presented here might serve this aim. We, however, acknowledge that our study is not suitable to provide accuracy values for each of the techniques recommended here. However, our work was meant to provide a practical framework, which, of course, relies also on our clinical experience, to aid the differential diagnosis of PED. We would advise applying the algorithm described here in every patient with PED, especially in those without additional symptoms or signs, before considering a genetic evaluation.

Author Roles

(1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript: A. Writing of the First Draft, B. Review and Critique.

R.E.: 1A–C, 3A–B

M.S.: 1C, 3B

C.G.: 1C, 3B

M.S.: 1C

V.H.:1C

A.B.:1C, 3B

K.B.: 1A–B, 3B

Financial Disclosures

R.E. is partly supported by COST Action BM1101 (reference: ECOST‐STSM‐BM1101‐160913‐035934). C.G.: academic research support: Deutsche Forschungsgemeinschaft (MU1692/2‐1 and GA 2031/1‐1) and European Science Foundation; commercial research support: travel grants by Actelion, Ipsen, Pharm Allergan, and Merz Pharmaceuticals. K.P.B. receives royalties from the publication of Oxford Specialist Handbook Of Parkinson's Disease and Other Movement Disorders (Oxford University Press, 2008) and of Marsden's Book of Movement Disorders Oxford (University Press, 2012); he received funding for travel from GlaxoSmithKline (GSK), Orion Corporation, Ipsen, and Merz Pharmaceuticals. M.S. receives academic research support from the Kapodistrian University of Athens and the European Union (reference: KA 70/3/11679 and MIS 377206) as well as travel and speaker honoraria from Actelion Pharmaceuticals. All other authors have no disclosures.

Supporting information

TABLE S1. Investigation results in our series

Video 1. Segment 1: case 8 showing PED of his left foot after having walked for approximately 2 miles. Segment 2A: case 7 with no abnormalities when initiating cycling. Segment 2B (slow motion): case 7 after 10 minutes of cycling. Segment 3A: case 16 showing mild spasticity of her right leg. Segment 3B: case 16 having an attack after having walked for 20 minutes.