Assessment of a Targeted Gene Panel for Identification of Genes Associated With Movement Disorders

Key Points

Question  Is a customized gene panel study suitable for the identification of genes associated with movement disorders?

Findings  This study aimed at developing a targeted sequencing approach using a panel of 127 genes involved in movement disorders and evaluating its performance through analysis of a cohort of 378 patients. A diagnostic yield of 22% was achieved, highlighting some unexpected genotype-phenotype correlations and 49 novel pathogenic variants.

Meaning  High-coverage sequencing panel to identify genes associated with movement disorders provided a useful and efficient diagnostic alternative to whole-exome and whole-genome sequencing.

Importance  Movement disorders are characterized by a marked genotypic and phenotypic heterogeneity, complicating diagnostic work in clinical practice and molecular diagnosis.

Objective  To develop and evaluate a targeted sequencing approach using a customized panel of genes involved in movement disorders.

Design, Setting and Participants  We selected 127 genes associated with movement disorders to create a customized enrichment in solution capture array. Targeted high-coverage sequencing was applied to DNA samples taken from 378 eligible patients at 1 Luxembourgian, 1 Algerian, and 25 French tertiary movement disorder centers between September 2014 and July 2016. Patients were suspected of having inherited movement disorders because of early onset, family history, and/or complex phenotypes. They were divided in 5 main movement disorder groups: parkinsonism, dystonia, chorea, paroxysmal movement disorder, and myoclonus. To compare approaches, 23 additional patients suspected of having inherited cerebellar ataxia were included, on whom whole-exome sequencing (WES) was done. Data analysis occurred from November 2015 to October 2016.

Main Outcomes and Measures  Percentages of individuals with positive diagnosis, variants of unknown significance, and negative cases; mutational frequencies and clinical phenotyping of genes associated with movement disorders.

Results  Of the 378 patients (of whom 208 were male [55.0%]), and with a median (range) age at disease onset of 31 (0-84) years, probable pathogenic variants were identified in 83 cases (22.0%): 46 patients with parkinsonism (55% of 83 patients), 21 patients (25.3%) with dystonia, 7 patients (8.4%) with chorea, 7 patients (8.4%) with paroxysmal movement disorders, and 2 patients (2.4%) with myoclonus as the predominant phenotype. Some genes were mutated in several cases in the cohort. Patients with pathogenic variants were significantly younger (median age, 27 years; interquartile range [IQR], 5-36 years]) than the patients without diagnosis (median age, 35 years; IQR, 15-46 years; P = .04). Diagnostic yield was significantly lower in patients with dystonia (21 of 135; 15.6%; P = .03) than in the overall cohort. Unexpected genotype-phenotype correlations in patients with pathogenic variants deviating from the classic phenotype were highlighted, and 49 novel probable pathogenic variants were identified. The WES analysis of the cohort of 23 patients with cerebellar ataxia led to an overall diagnostic yield of 26%, similar to panel analysis but at a cost 6 to 7 times greater.

Conclusions and Relevance  High-coverage sequencing panel for the delineation of genes associated with movement disorders was efficient and provided a cost-effective diagnostic alternative to whole-exome and whole-genome sequencing.

Movement disorders are a set of heterogeneous neurological syndromes affecting the ability to produce and control movement because of dysfunction in the basal ganglia and/or connected structures.

Movement disorders can be acquired or can arise because of numerous inherited diseases, which are characterized by a great clinical and genetic heterogeneity and a frequent clinical overlap, which prevents reliable genotype-phenotype correlations. The diagnostic work in clinical practice remains a challenging issue, and molecular diagnosis by standard Sanger sequencing is tedious, time consuming, and inefficient, particularly because analysis of the known implicated genes are not yet always routinely available.

The aim of this study was to develop a genetic diagnostic strategy based on a high throughput sequencing technology targeting 127 genes involved in movement disorders. This work intended to assess the efficiency of this approach as a diagnostic tool, help to define mutational frequencies and phenotypic spectra, bring additional evidence for the associations of previous candidate genes with diseases, identify novel mutations, and improve genotype-phenotype correlations.

In the multicentric prospective study, patients were selected from 1 Luxembourgian, 1 Algerian, and 25 French tertiary centers specializing in the treatment of movement disorders. Inclusion criteria were that patients (1) had developed 1 or more chronic movement disorders and (2) had an age at onset younger than 40 years and/or a family history of movement disorders. Patients with essential tremor, tic or Tourette syndrome, pure cerebellar ataxia, or clinical or paraclinical findings suggestive of an acquired cause were not included. Most of the patients underwent common biochemical testing (eg, for copper and ceruloplasmin), brain imaging, and most common relevant genetic tests (such as testing for Wilson disease or Huntington disease) before their inclusion. Demographic, clinical, and paraclinical data were collected, as well as family history and a family tree.

The protocol for this study was approved by the Hôpitaux Universitaires de Strasbourg institutional review board. For all patients, a written informed consent for genetic testing was obtained; in the case of minor participants, this was either from adult probands or from a legal representative.

In the current nonautomated setting of this study, we manually processed each series of 24 DNA samples corresponding to index cases (or series of 6 exomes of index cases) for library preparation, and then libraries were pooled for capture and enrichment reaction. For the sequencing step, we used 1 lane of the Hiseq 4000 Sequencing System (Illumina) flow cell either for the targeted sequencing of 24 samples or the WES of 6 samples. The final step corresponded to the bioinformatics analysis of sequencing data, the interpretation of the variants, and the writing of diagnostic reports. The 127 genes included in the panel are listed in eTable 1 in the Supplement.

Analysis included identification of variants of unknown significance (VUSs). Novel variants were considered as VUSs when they were rare (minor allele frequency < 0.1%) in population databases (eg, Exome Variant Server, Exome Aggregation Consortium, and 1000 Genomes Browser) and/or predicted to be pathogenic by prediction tools (eg, Sorting Intolerant From Tolerant [SIFT], PolyPhen, and MutationTaster), but when the phenotype expressed by the patient was not consistent with the usual phenotype described in the literature.

The diagnostic yield of the movement disorder gene panel was compared with the diagnostic yield of whole-exome sequencing (WES) by performing WES in 23 patients suspected of having autosomal recessive cerebellar ataxia because of early age at onset and/or consanguinity. This analysis was combined with that of previously published WES data on inherited cerebellar ataxias in a sample of 76 patients.¹

Comparisons between 2 groups were tested for statistical significance using a χ² test for qualitative variables and a t test for quantitative variables. The Fisher exact test was used in small groups. We considered P values of .05 or less statistically significant. For genetic analyses, see eMethods in the Supplement.

Data analyses were completed with the Polyweb interface (Université Paris Descartes and Institut Imagine). Data analysis took place from November 2015 to October 2016.

A total of 378 patients were included in the study, of whom 208 male individuals (55.0%) and 170 female individuals (45.0%). The median age at onset (MAO) of the disease was 31 years (range, 0-84 years). The median age at study inclusion was 45 years (range, 0-87 years). At least 144 patients (38.0%) had undergone genetic testing before their inclusion in the study.

A total of 238 of the 378 patients in the cohort (63.0%) had sporadic cases, and the remaining 140 (37.0%) had a family history of disease. Patients were classified according to their prominent movement disorders: parkinsonism (n = 181 of 378 [47.9%]), dystonia (n = 135 [35.7%]), chorea (n = 25 [6.6%]), paroxysmal movement disorders (n = 20 [5.3%]), and myoclonus (n = 17 [4.5%]). Another 166 patients (43.9%) presented with combined movement disorders; this included 12 of the 17 patients with myoclonus (70.6%) and 16 of the 25 patients with chorea (64.0%), as well as 65 of the 135 patients with dystonia (48.1%) and 71 of the 181 patients with parkinsonism (39.2%). Rest tremor was not considered an independent clinical feature in patients with parkinsonism. Associated clinical features are detailed in Figure 1.

Our strategy allowed generating a high-quality sequencing data set, with a mean depth of coverage of 1266 × and a mean of 99.7% of targeted regions being well covered in each patient (× >30).² We systematically validated relevant variants identified in the first 48 patients of the cohort and found perfect coherency. Moreover, previously known mutations in the 3 control patients were found by the method. Interestingly, the pipeline seemed to be efficient in copy number variation calling.

We detected pathogenic variants in 83 of 378 patients (Table; Figure 2, Figure 3; eFigure 1 in the Supplement), leading to an overall diagnostic yield of 22.0%. Diagnosis was less frequently established in the dystonia group (n = 21 of 135; 15.6%; P = .03) compared with the parkinsonism group (n = 46 of 181; 25.4%; P = .04) and the rest of the cohort (n = 62 of 243; 25.5%; P = .02). In the 83 patients with genetic diagnoses, the MAO was significantly lower (median, 27 years [IQR, 5-36 years]) than the patients without clear molecular diagnosis (median, 35 years [IQR, 15-46 years]; P = .04). In patients with genetic diagnosis, a family history was more frequently found (patients with genetic diagnosis: n = 36 [43.4%] vs patients without genetic diagnosis: n = 103 [34.9%]), as was male predominance (men: n = 50 [60.2%] vs women: n = 156 [52.9%]); however, these comparisons were but without statistical significance. A total of 49 novel probable pathogenic variants were identified.

Seventeen patients harbored variants in PARKIN, even though it had already been tested and found negative in 38 members (10.0%) of the entire cohort before inclusion in this study. In addition, 9 patients had variants in GBA, and another 8 patients had variants in LRRK2. Numerous other very rare entities (with an estimated minor allele frequency of < 0.01%) were identified, such as PLA2G6 (n = 4), VPS13A (n = 4), ATP13A2 (n = 3), DNAJC13 (n = 3), GCH1 (n = 3), ADCY5 (n = 2), AP4B1 (n = 2), DJ-1 (n = 2), GLRA1 (n = 2), VPS35 (n = 2), and WDR45 (n = 2). The rate of identified pathogenic variants was higher in paroxysmal movement disorders (n = 7 of 20; 35%), chorea (n = 7 of 25; 28%), and parkinsonism (n = 46 of 181; 25.4%) than in the cohort overall (n = 83 of 378; 22.0%).

The WES analysis on a cohort of 23 patients with predominantly recessive cerebellar ataxia led to an overall diagnostic yield in 10 patients (43.5%; eTable 2 and eTable 3 in the Supplement). To get significant diagnostic yield for the WES study, our results were combined with the published series of Fogel et al,¹ who obtained a yield of diagnostic pathogenic gene variants in 16 patients in a sample of 76 (21.1%) for a similar series of cerebellar ataxia patients and WES analysis strategy. The overall diagnostic yield for pathogenic gene variants by WES in cerebellar ataxia is therefore 26 of 99 (26.3%), which is similar to the panel approach for movement disorders.

No relevant variants were found in 244 patients (64.6%), with no significant difference between phenotype groups (Figure 2). Their characteristics were similar to the initial cohort, with an MAO of 30 years (IQR, 12-44 years), a proportion of 54.1% male probands (n = 132 of 244), and a proportion of 38.1% familial cases (n = 93 of 244), with family trees suggesting mendelian inheritance (eFigure 2 in the Supplement). These characteristics were not statistically different than the initial cohort.

A total of 74 VUSs, which have implications in movement disorder that remain doubtful and must be further studied, were identified in 60 other patients (15.9%; eTable 4 and eFigure 3 in the Supplement; Figure 2). Of the 74 VUSs, 69 (93.2%) were identified in an heterozygous or hemizygous state, and 5 (6.8%) in a homozygous or compound heterozygous state. The age at disease onset, sex, and proportion of family history in patients with VUSs did not significantly differ from the overall cohort.

In 4 patients, the frameshift deletion c.101_102delAG; Q34Gfs*5 (34Q−) was identified in PARKIN in a homozygous or compound heterozygous state. In this cohort, when found at compound heterozygous state, it was associated either with a juvenile-onset dystonia-parkinsonism phenotype or with a typical Parkinson disease presentation. Interestingly, in 2 patients (consanguineous siblings of Algerian descent) who were harboring the homozygous variant, the only symptom was an exercise-induced transient lower-limb dystonia, 10 years after disease onset.

We found pathogenic variants in 46 of 181 patients (25.4%) with parkinsonism. The MAO in patients with heterozygous variants was 44 years, vs 27 years in patients with homozygous or compound heterozygous variants. In our cohort, LRRK2 variants were identified in 9 patients, including 7 G2019S mutations, 1 L1114L variant (described as pathogenic by Zimprich at al⁴), and 1 novel L448S variant (which was deemed probably causative). All patients were presenting with typical Parkinson disease. Interestingly, the MAO was less than 40 years in 6 patients, and cognitive impairment was found in 1 patient. In this study, 8 patients had heterozygous GBA variations, 5 of whom developed typical Parkinson disease, with an age at onset of less than 40 years in 3 patients. Among these 5 patients with Parkinson disease, 2 had either the L483P or the R502H variants, which have already been identified as risk factors for Parkinson disease, and 3 had novel variants: R398*, W223R, or S235P. Two heterogenous missense mutations were identified in DNAJC13, in 1 patient experiencing Parkinson disease and cervical dystonia with an age at onset of 60 years (variant R1165G) and in 2 affected siblings with Parkinson disease with ages at onset of 56 and 60 years (variant K925Q). One missense mutation of HTRA2 has been identified in 1 patient with typical Parkinson disease. Interestingly, 1 hemizygous somatic mosaic point mutation was found in WDR45 in a male patient with intellectual disability, sporadic early-onset parkinsonism, and T2-weighted magnetic resonance imaging results showing a hyposignal within the substantia nigra. Variants in this gene have been reported as a cause of an X-linked dominant neurodegeneration with brain iron accumulation characterized by childhood developmental disability followed by adolescent or adult onset of dystonia, parkinsonism, and dementia⁵ (eResults in the Supplement). These results supported the implication of DNAJC13 and HTRA2 in Parkinson disease.

In patients with dystonia, causative variants were identified in only 21 of 135 cases (15.6%). We found 2 consanguineous sisters with childhood-onset myoclonus dystonia harboring homozygous pathogenic variant P222L in PRKRA, a gene that has been previously identified in very rare recessive early-onset progressive limb dystonia, laryngeal and oromandibular dystonia, or dystonia-parkinsonism.⁶ Unexpectedly, we identified a pathogenic variant in MECP2 in a 50-year-old female patient who had presented with an undiagnosed mild form of Rett syndrome combined with cerebellar ataxia and dystonic tremor.⁷ In addition, a 19-year-old patient was diagnosed with a mild form of sulfite oxidase deficiency (SUOX),⁸ and an SLC2A1 variant was found in 1 patient, aged 29 years, who was presenting with delayed psychomotor development, intellectual disability, atonic absence seizures, permanent axial and lower limb dystonia, pyramidal signs, action myoclonus, and microcephaly (eResults in the Supplement).

Regarding cervical dystonia, no relevant variants were identified in genes that were recently identified in WES-based studies (for instance, CIZ1 and ANO3).^9,10 However, 1 patient harbored a variant in TUBB4A, and a second patient harbored a variant in GCH1.

Considering chorea, probable pathogenic variants were found in 7 of 25 patients (28%), in ADCY5, ALDH5A1, VPS13A and XK, which are known to be involved in chorea. In the study cohort, the 2 patients harboring ADCY5 variants presented with infantile-onset chorea, facial myokymia, and action myoclonus in 1 case and with childhood-onset generalized dystonia associated with action myoclonus but also with vertical gaze palsy in the second case. In this case, the presentation initially led clinicians to suspect Niemann-Pick disease type C.

Strikingly, causative variants were identified in 7 of 20 patients (35%) with paroxysmal movement disorders. Homozygous AP4B1 variants were found in 2 sisters with infantile-onset developmental delay, epilepsy, paroxysmal movement disorders (eg, myoclonus), and spastic paraplegia associated with dysmorphic features and a thin corpus callosum.

In 2 of the 17 patients (12%) who presented with myoclonus as a predominant movement disorder, variants were found in ATP1A3 and ATP13A2. Our findings also emphasize marked genetic and phenotypical heterogeneities, as well as the overlaps in the field of movement disorders that were highlighted in this series of findings (Figure 4).

Several other probable pathogenic variants were identified in this cohort, especially R393* in AP4B1 and K925Q in DNAJC13, which were found respectively in 2 affected members of a single family. Segregation analysis supported the implication of the variant in AP4B1 in the 2 affected siblings, but it could not be performed in the family with the DNAJC13 variant because of the unavailability of the parents’ DNA.

In this study, deletions in the exon 25 of GIGYF2 (the pathogenicity of which is debated^11,12) were identified in 4 patients (considered negative results; data not shown) who presented very different phenotypes. Three patients were without any parkinsonian feature, with 2 patients presenting instead with generalized dystonia and 1 patient presenting with myoclonus. Relevant pathogenic variants were found in only 1 of 35 patients (3%) presenting with an associated cerebellar ataxia.

The estimated cost of the gene panel strategy without labor costs was $156 per patient (reagent cost of $70, run cost of $68, and $18 of consumable items), whereas the cost of WES without labor costs was estimated between $850 and $1113 per patient. The amount of time required for the gene panel strategy for a series of 24 patients varied between 2 to 3 full days, while data analysis of a series of 6 exomes required 7 to 10 days.

Targeted sequencing of 127 genes in a cohort of 378 patients of mostly European descent with unknown etiology of movement disorders led to a conclusive diagnostic yield of 22.0%. This rate of success is in good accordance with other studies using targeted high throughput sequencing in the diagnosis of neurological diseases.^13,14 The cost-effectiveness of gene panel analyses has already been shown in the field of movement disorders.¹⁴ In our series, the rate of molecular diagnosis was significantly higher when the MAO was lower. The high depth of coverage and the smaller portion of poorly covered regions achieved with our strategy ensure a high sensitivity and specificity of detecting pathogenic events in the regions of interest and enable the identification of single-nucleotide variants, indels, but also copy number variants, which remain a challenge in next-generation sequencing.¹⁵ It also showed accuracy for the detection of somatic mutations. However, it remains incapable of detecting triplet repeat expansions, therefore the search for Huntington disease; spinocerebellar ataxia types 1, 2, 3, 6 and 7; fragile X-associated tremor ataxia syndrome; or dentatorubral-pallidoluysian atrophy was impossible. Furthermore, variant ranking could fail to detect relevant variants with incomplete penetrance because of a higher allele frequency in the general population.

The variants in many genes identified in our series were responsible for complex pictures with combined movement disorders. These findings reinforce the interest to use a gene panel devoted to several distinct types of movement disorders, beyond dystonia and parkinsonism, which were the most frequent signs in our cohort.

Despite the relatively small size of the cohort, this study gives some interesting epidemiological data, including confirmation of the 3 genes most frequent involved in Parkinson disease¹⁶ and the suggestion that they should be more systematically searched for. Less expectedly, numerous other very rare entities were identified. Variants in ALDH5A1, ATP1A3, DJ-1, DNAJC13, GBA, GCH1, LRRK2, MECP2, NKX2-1, PINK1, PLA2G6, PRKRA, PRRT2, SLC2A1, SUOX, VPS13A, and VPS35 could cause a complex picture with several movement disorders. Variants in ADCY5, ATP13A2, and PARKIN could be responsible for a wider spectrum of phenotypes, with different possible prominent MDs. By contrast, a few genes were associated with a relatively pure phenotype, as EIF4G1 (Parkinson disease), GNAL, TOR1A and TUBB4A (dystonia), XK (chorea), PRRT2 (paroxysmal kinesigenic dyskinesia), and GLRA1 (hyperekplexia).

To our knowledge, the 34Q− variant of PARKIN had never been described in patients with movement disorders before now. Exercise-induced dystonia has already been described as a presenting feature of young-onset Parkinson disease,¹⁷ similar to the phenotype seen in 2 patients in this study, but other signs of the disease usually appeared within the 5 following years, as well as early-onset levodopa-induced dyskinesias, in contrast to their lack of symptoms 10 years after onset.

It is remarkable that the diagnostic yield in choreic patients was quite high (28%), although Huntington disease, which is known to be the most frequent inherited cause of chorea, could not be diagnosed with the gene panel. These results supported the implication of DNAJC13 and HTRA2 in Parkinson disease, 2 genes whose implication has recently been reported and remains debated.^18,19

Interestingly, the rate of identified pathogenic variants was significantly lower in dystonia (16%), although most of the targeted genes (69 of 127) were associated with dystonia as the prominent clinical sign of disease. The data suggest that PRKRA should be added to the genes tested in patients with myoclonus dystonia. With respect to cervical dystonia, 1 patient had had a good response to levodopa, suggesting that a trial dopaminergic treatment might be tested in patients with isolated cervical dystonia. In addition, neurological affections which are more frequently encountered in pediatric patients were identified in 3 adults with dystonia: sulfite oxidase deficiency (SUOX), Rett syndrome (MECP2), and GLUT1 deficiency syndrome (SCL2A1). These results highlight the fact that these diagnoses should be considered in adult patients. In particular, SLC2A1 mutations should be searched for in patients with complex movement disorders even when the symptoms are permanent and without exacerbation after fasting, since ketogenic diet is highly effective in reducing clinical features especially when started early in childhood. Moreover, for all these patients, even when there is no specific treatment, a suitable genetic counseling can be provided to the patients and their relatives. A lower diagnostic yield was found in patients displaying myoclonus as main sign of the disease, because of the small number of ataxia genes included in the panel (approximately 10 of 127 included genes).

The occurrence of deletions in the exon 25 of GIGYF2 in 3 patients without parkinsonian clinical features is evidence against any specific role of these variants in Parkinson disease. By contrast, several heterozygous pathogenic variants were identified in PARKIN in patients with Parkinson disease, supporting the previous assertion that heterozygous pathogenic variants are a genetic susceptibility factor for Parkinson disease.²⁰ Several copy number variants, which are frequently found in the heterozygous state in PARKIN-type of early-onset Parkinson disease²¹ were also identified in these patients, as well as missense mutations, with a higher mutational frequency than in population databases. The MAO was consistent with literature data.²² Strikingly, L444P and N370S, the 2 most frequent mutations in previous studies, were not found in this cohort.^3,23-25

However, 8 VUSs were identified in LRRK2. Further studies will be necessary in order to assess whether they are pathogenic.

Though a WES approach enables to explore all genes, WES analysis costs up to 6 to 7 times more than the panel analysis, in addition to being more time-consuming and more likely to lead to the identification of a large number of VUSs. For these reasons, panel analysis seems to be an acceptable alternative for diagnostic setting. The targeted sequencing of 127 genes in this cohort of 378 patients with unknown genetic etiology of movement disorders led to a conclusive diagnostic yield of approximately 25%, which is similar to the overall diagnostic rate by WES for cerebellar ataxia (26%). When restricted to patients with recessive or sporadic cerebellar ataxia and age at onset younger than 20 years, the detection rate increased to 45% in this series (10 of 22), which was similar to the results of Fogel et al¹ for this class of patients (8 of 15 patients [53%]). It therefore appears that efficiency of WES and targeted high-coverage sequencing highly depends on the inclusion criteria, and that broad inclusion criteria that are required in a diagnostic setting result in reduced rate. Moreover, caution should be used for interpretation of the difference between diagnostic yields, as recently reported WES-based studies on diverse heterogeneous groups of disorders revealed detection rates for confirmed pathogenic mutations ranging from 21% to 41%.^26,27

Because of the complex genetic and phenotypic heterogeneity of movement disorders, precise molecular diagnosis remains rarely proposed for most patients in France. It is therefore timely to replace direct sequencing strategies consisting in sequentially analyzing candidate genes by more high-throughput next-generation-sequencing–based strategies, such as targeted sequencing of up to several hundred of genes, or WES. Strategies based on WES are very attractive because of their possible application whatever the group of disorders and the clinical features. However, in diagnostic settings, mean coverage and depth usually provided by WES are often considered insufficient for detection of mosaic mutations. Also, it is well established that a WES approach frequently leads to the identification of numerous VUSs or conflicting interpretations. Such findings usually lead to time-consuming research and literature review activity to structure objective opinions and interpretations that are not necessarily translated and delivered to concerned families as diagnostic results.

For these reasons, targeted sequencing appears more appealing for diagnostic settings, and with an even broader panel of genes that includes (for instance) genes involved in ataxia and spastic paraplegia, a higher resulting diagnostic yield could be expected. The high depth of coverage and the smaller portion of poorly covered regions achieved with our strategy ensure a high sensitivity and specificity of detecting pathogenic events in the regions of interest (single-nucleotide variants, indels, but also copy number variants), an issue that remains difficult to address in WES-based studies. Nonetheless, one of the limitations of targeted sequencing is that analysis of newly identified genes is difficult unless diagnostic laboratory can afford regular revision and implementation of renewed and adapted panels. Moreover, targeted sequencing of movement disorder–associated genes offers the possibility of applying such tests to a high proportion of patients awaiting molecular diagnosis, given the significantly lower cost of sequencing, amount of time dedicated for data analysis and interpretation.

With respect to the cost of targeted high-coverage sequencing of movement disorder–associated genes, continuous efforts are being made to optimize the procedure and reduce costs while preserving the quality. In our current setting, consumable for the processing and sequencing of 1 DNA sample is as low as $120, while WES cost varies between $834 to $1200.

The amount of time required for WES strategy was 3 to 4 times higher than for the gene panel strategy, and the panel analysis is moreover less complicated. Of course, the amount of time dedicated to data analysis could be significantly reduced if a WES approach is applied for trios of patients and parents. However, this strategy is often hampered by the availability of the DNA of the parents and by significant additional cost, limiting its use. After reevaluation of clinical and familial criteria (eg, age at disease onset, severity, and family history), we believe that 38 of 244 patients for whom no relevant mutation was identified would need to be explored by WES, preferably through a trio strategy.

This study had several limitations. Most patients were European in origin; however, even if the patients included in the present study were from multiple racial/ethnic origins, one would have to be careful when extrapolating the results to other populations because of variability of mutation frequencies and founder effects. In addition, this study is subject to biases because of the small number of patients in each group, the great variability in previous analyses and data collection, the questionable choice of the main movement disorder in many patients, and the fact that dystonia was a frequent cause of inclusion.

As in many other studies based on targeted approaches,²⁸ the substantial proportion of patients of our investigated cohort that remain without molecular diagnosis raises several hypotheses. Among them, there are the hypothesis of potential additional movement disorder–associated genes that remain to be discovered, the possibility of missed nonexplored mutations located in noncoding regions, and lastly whether more complex genetic patterns involving for example variants with reduced penetrance and oligogenic and/or multifactorial modes of inheritance.^29-31

In conclusion, this high-throughput sequencing strategy targeting 127 genes appears as a highly efficient, cost-effective diagnostic tool in the field of early-onset or familial movement disorders. Given the inconstant genotype-phenotype correlations, the frequency of combined movement disorders, and the overlaps, the relevance of targeted sequencing as a first intention test for the diagnosis of movement disorders should be considered. It would be interesting to investigate a larger cohort and to improve the design of the targeted regions, especially adding more genes causing cerebellar ataxia, and complete the study by WES analysis on unresolved cases to search for variants in recently identified new genes or in novel candidate genes.

Corresponding Author: Mathieu Anheim, MD, PhD, Département de Neurologie, Hôpital de Hautepierre, 1 Ave Molière, 67098 Strasbourg Cedex, France (mathieu.anheim@chru-strasbourg.fr).

Accepted for Publication: March 8, 2018.

Published Online: June 18, 2018. doi:10.1001/jamaneurol.2018.1478

Author Contributions: Drs Anheim and Chelly had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Montaut and Tranchant equally contributed to the work and should be considered co-first authors. Drs Chelly and Anheim equally contributed to the work and should be considered co-last authors.

Study concept and design: Montaut, Tranchant, Drouot, Mandel, Chelly, Anheim.

Acquisition, analysis, or interpretation of data: Montaut, Tranchant, Drouot, Rudolf, Guissart, Tarabeux, Stemmelen, Velt, Fourrage, Nitschké, Gérard, Koenig, Chelly, Anheim.

Drafting of the manuscript: Montaut, Rudolf, Tarabeux, Fourrage, Chelly, Anheim.

Critical revision of the manuscript for important intellectual content: Tranchant, Drouot, Guissart, Stemmelen, Velt, Nitschké, Gérard, Mandel, Koenig, Chelly, Anheim.

Statistical analysis: Montaut.

Obtained funding: Mandel, Chelly, Anheim.

Administrative, technical, or material support: Montaut, Drouot, Rudolf, Tarabeux, Gérard, Chelly, Anheim.

Study supervision: Montaut, Tranchant, Chelly, Anheim.

Conflict of Interest Disclosures: None reported.

Funding/Support: The study has been supported by a grant from France Parkinson.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The French Parkinson’s and Movement Disorders (PMD) Consortium: Département de Neuropédiatrie, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg, Strasbourg, France: Marie-Thérèse Abi-Warde, MD; Anne De Saint Martin, MD. Département de Neurologie, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg, Strasbourg, France: Charles Behr, MD, PhD; Ouhaid Lagha Boukbiza, MD. Département de Neurologie, Centre Hospitalier Universitaire de Marseille, Hôpital de la Timone, Marseille, France: Jean-Philippe Azulay, MD, PhD; Emmanuelle Boutin, MD; Alexandre Eusebio, MD, PhD; Frédérique Fluchere, MD; Elsa Kaphan, MD. Département de Neurologie, Centre Hospitalier Universitaire Henri Mondor, Créteil, France: Anne Catherine Bachoud Levi, MD, PhD; Gilles Fenelon, MD, PhD; Jean-Marie Gurruchaga, MD. Service de Neurologie, Hôpital de Mercy, Centre Hospitalier de Metz-Thionville, Metz, France: Carmen Badiu, MD. Service de Neurologie, Centre Hospitalier de la Côte Basque, Bayonne, France: Charles Behr, MD. Département de Neurologie, Centre Hospitalier Régional Universitaire de Besançon, Besançon, France: Matthieu Bereau, MD; Ludivine Chamard, MD. Département de Neurologie, Centre Hospitalier Universitaire de Toulouse, Toulouse, France: Christine Brefel-Courbon, MD, PhD; Jonathan Curot, MD; Julia Dupouy, MD. Département de Neurologie, Centre Hospitalier Universitaire de Bordeaux, Bordeaux, France: Pierre Burbaud, MD, PhD; Virginie Chrystostome, MD; Antoine Gradel, MD. Département de Neurologie, Centre Hospitalier Universitaire de Grenoble, France: Anna Castrioto, MD, PhD. Département de Neurochirurgie, Centre Hospitalier Universitaire de Montpellier, Montpellier, France: Ana Laura Cif, France, MD, PhD; Victoria Gonzales, MD. Département de Neurologie, Centre Hospitalier Universitaire Paris–GH La Pitié Salpêtrière–Charles Foix–Hôpital Pitié-Salpêtrière, Assistance Publique–Hôpitaux de Paris, Paris, France: Maria Clara Coelho Braga, MD; Pauline Dodet, MD; Bertrand Degos, MD, PhD; Emmanuel Flamand-Roze, MD, PhD; David Grabli, MD, PhD; Andreas Hartmann, MD, PhD. Département de Neurologie, Centre Hospitalier Universitaire de Nantes, Nantes, France: Philippe Damier, MD, PhD, MBA; Pascal Derkinderen, MD, PhD. Département de Neurologie, Centre hospitalier universitaire de Reims, Reims, France: Anne Doe De Maindreville, MD. Département de Neuropédiatrie, Centre Hospitalier Universitaire Paris Est–Hôpital d'Enfants Armand-Trousseau, Assistance Publique–Hôpitaux de Paris, Paris, France: Diane Doummar, MD, PhD. Département de Neurologie, Centre Hospitalier Universitaire de Rennes, Rennes, France: Sophie Drapier, MD. Service de Génétique Clinique, Pôle de Pédiatrie Médico-Chirurgicale Et Génétique Clinique, Centre Hospitalier Universitaire de Rennes–Hôpital Sud, Rennes, France: Mélanie Fradin, MD. Département de Neurologie, Centre Hospitalier Universitaire de Nancy, Nancy, France: Solène Frismand, MD. Service de Neurologie, Centre Hospitalier Universitaire Gabriel-Montpied, Clermont-Ferrand, France: François Gampourou, MD. Département de Neuropédiatrie, Hôpital Universitaire Robert Debré, Assistance Publique–Hôpitaux de Paris, Paris: Domitille Gras, MD. Service de Génétique/Unité de Génétique Clinique, Centre Hospitalier Universitaire de Rouen–Hôpital Charles Nicolle, Rouen, France: Lucie Guyant-Marechal, MD. Fondation Ophtalmologique Adolphe de Rothschild, Paris, France: Cécile Hubsch, MD, PhD. Neuropédiatrie, Hôpital Pellegrin-Enfants, Centre Hospitalier Universitaire de Bordeaux, Bordeaux, France: Marie Husson, MD. Département de Neurologie, Centre Hospitalier de Luxembourg, Luxembourg: Rejko Krüger, MD. Service de Neurologie, Centre Hospitalier Universitaire Amiens-Picardie–Site Sud, Amiens, France: Pierre Krystkowiak, MD, PhD. Département de Neurologie, Hôpital Pierre Wertheimer, Hôpitaux Civils de Lyon, Lyon, France: Chloé Laurencin, MD. Département de Neurologie, CHU de Bordeaux, Bordeaux, France: Brice Laurens, MD. Département de Neurologie, CHU de Rouen - Hôpital Charles Nicolle, Rouen, France: Romain Lefaucheur, MD; Gaël Nicolas, MD, PhD; Marie-Laure Welter, MD, PhD. Service de Neurologie Pédiatrique, Hôpital Femme Mère Enfant, Hôpitaux Civils de Lyon, Lyon, France: Laurence Lion-Francois, MD. Département de Neurologie, Hôpital Nord - CHU Grenoble Alpes, Grenoble, France: Martial Mallaret, MD, PhD. Département de Neurologie, CHU Paris-GH La Pitié Salpêtrière-Charles Foix - Hôpital Pitié-Salpêtrière, AP-HP, Paris, France: Louise-Laure Mariani, MD; Yann Nadjar, MD; Savine Vicart, MD; Marie Vidailhet, MD, PhD. Pôle des Neurosciences Cliniques, CHU de Bordeaux-GH Pellegrin, Bordeaux, France: Wassilios Meissner, MD, PhD. Service de génétique, CHU de Reims - Hôpital Maison Blanche, Reims, France: Monique Mozelle Nivoix, MD. Département de Neurologie, CHU de Toulouse, Toulouse, France: Fabienne Ory-Magne, MD; Olivier Rascol, MD, PhD. Centre Hospitalier Sud-Francilien, Corbeil-Essones, France: Ivania Patry, MD. Département de Neurologie, Hôpital Pierre Wertheimer, Hôpitaux Civils de Lyon, Lyon, France: Alice Poisson, MD, PhD. Département de Neurologie, CHU Henri Mondor, Créteil, France: Philippe Remy, MD, PhD; Hayet Salhi, MD. Département de Neurologie, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg, Strasbourg, France: Mathilde Renaud, MD, PhD. Département de Neurologie, CHU de Marseille - Hôpital de la Timone, Marseille, France: Wissam Rizk, MD; Tatiana Witjas, MD. Département de Neurologie, Hôpital G.R. Laennec, CHU de Nantes, Nantes, France: Tiphaine Rouaud, MD. Département de neuropédiatrie, Pôle Femme Mère Enfant, CHRU de Montpellier - Hôpital Gui de Chauliac, Montpellier, France: Agathe Roubertie, MD, PhD. Service de Neurologie, Hopital Zobda-Quitman, CHU de Fort-de-France, Martinique, France: Aissatou Signate, MD. Département de Neuropédiatrie, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg, Strasbourg, France: Marie-Aude Spitz, MD. Département de Neurologie, Centre hospitalier universitaire Le Kremlin Bicêtre, AP-HP, Paris, France: Maeva Stephant, MD. Service de Neurologie, CHU Amiens-Picardie - Site Sud, Amiens, France: Mélissa Tir, MD, PhD. Service de Neurologie, CHU de Bordeaux-GH Pellegrin, Bordeaux, France: François Tison, MD, PhD. Service de Neurologie, CHU Annaba, Annaba, Algeria: Nadia Toubal, MD. Service de Neurologie, CHU Gabriel-Montpied, Clermont-Ferrand, France: Miguel Ulla, MD, PhD. Département de Neurologie, CHU de Rennes, Rennes, France: Marc Verin, MD, PhD. Service de Neurologie, Centre Hospitalier de Verdun, Verdun, France: Jean-Médard Zola, MD.

Additional Contributions: The authors thank the patients and their families for participation in this study, Bernard Jost, PhD, the IGBMC Microarray and Sequencing Platform; Cécilia Marelli, MD, PhD, the Department of Neurology of the University Hospital of Montpellier; Bernard Michel, MD, the Department of Neurology of the University Hospital of Marseille; Abderrahim M'zahem, MD, the University Hospital of Constantine; and Jean-François Deleuze, PhD, the Centre National de Génotypage (CNG) and the IdEx project for their support with design and conduct of the study. They received no compensation from a funding source for their contributions.