Genetic Diversity and Expanded Phenotypes in Dystonia: Insights from Large-Scale Exome Sequencing

Abstract

Dystonia is one of the most prevalent movement disorders, characterized by significant clinical and etiological heterogeneity. Despite considerable heritability (~25%) and the identification of several disease-linked genes, the etiology in most patients remains elusive. Moreover, understanding the correlations between clinical manifestation and genetic variants has become increasingly complex. To comprehensively unravel dystonia's genetic spectrum, we performed exome sequencing on 1,924 dystonia patients [40.3% male, 92.9% White, 93.2% isolated dystonia, median age at onset (AAO) 33 years], including 1,895 index patients, who were previously genetically unsolved. The sample was mainly based on two dystonia registries (DysTract and the Dystonia Coalition). Further, 72 additional patients of Asian ethnicity, mainly from Malaysia, were also included. We prioritized patients with negative genetic prescreening, early AAO, positive family history, and multisite involvement of dystonia. Rare variants in genes previously linked to dystonia (n=405) were examined. Variants were confirmed via Sanger sequencing, and segregation analysis was performed when possible. We identified 137 distinct likely pathogenic or pathogenic variants (according to ACMG criteria) across 51 genes in 163/1,924 patients [42.9% male, 85.9% White, 68.7% isolated dystonia, median AAO 19 years]. This included 153/1,895 index patients, resulting in a diagnostic yield of 8.1%. Notably, 77/137 (56.2%) of these variants were novel, with recurrent variants in EIF2AK2, VPS16, KCNMA1, and SLC2A1, and novel variant types such as two splice site variants in KMT2B, supported by functional evidence. Additionally, 321 index patients (16.9%) harbored variants of uncertain significance in 102 genes. The most frequently implicated genes included VPS16, THAP1, GCH1, SGCE, GNAL, and KMT2B. Presumably pathogenic variants in less well-established dystonia genes were also found, including KCNMA1, KIF1A, and ZMYND11. At least six variants (in ADCY5, GNB1, IR2BPL, KCNN2, KMT2B, and VPS16) occurred de novo, supporting pathogenicity. ROC curve analysis indicated that AAO and the presence of generalized dystonia were the strongest predictors of a genetic diagnosis, with diagnostic yields of 28.6% in patients with generalized dystonia and 20.4% in those with AAO < 30 years. This study provides a comprehensive examination of the genetic landscape of dystonia, revealing valuable insights into the frequency of dystonia-linked genes and their associated phenotypes. It underscores the utility of exome sequencing in establishing diagnoses within this heterogeneous condition. Despite prescreening, presumably pathogenic variants were identified in almost 10% of patients. Our findings reaffirm several dystonia candidate genes and expand the phenotypic spectrum of some of these genes to include prominent, sometimes isolated dystonia.

Keywords: Dystonia; Exome sequencing; Genetic heterogeneity; Pathogenic variants.

Figure 1. Overview of exome sequencing results.

(A) Proportion of index patients for whom a diagnostic variant was identified, who carry a variant of uncertain significance (VUS), and who remain unsolved after searching for pathogenic variants in dystonia-linked genes. (B) Proportion of diagnostic variants that were previously reported (known variant), not previously reported and detected in a single patient or pedigree (novel variant single), and not previously reported and found recurrently in at least two unrelated dystonia patients (novel variant recurrent). (C) Distribution of diagnostic variants by variant type.

Figure 2. Genetic landscape in our dystonia sample (n=1,895 index patients) identified by exome sequencing.

The number of individuals harboring a presumably pathogenic variant in genes previously linked to dystonia is shown, representing 51 distinct genetic forms and a total diagnostic yield of 8.1%. Truncating variants include stop-gain and frameshift variants. All variants were found in the heterozygous state and affect genes known to be associated with dominant inheritance, except for two homozygous variants in GCH1, one homozygous variant in SPR and SETX, respectively, and one hemizygous variant in MECP2.

Figure 3. Variants of uncertain significance in our dystonia sample (n=1,895 index patients) identified by exome sequencing.

The number of individuals harboring variants of uncertain significance in genes previously linked to dystonia is shown. This represents 329 distinct variants in 102 genes found in 321 index patients (16.9%). Truncating variants include stop-gain and frameshift variants. All variants were found in the heterozygous state and affect genes associated with dominant inheritance, except for one hemizygous variant in MECP2 and PDHA1, respectively.

Figure 4. Diagnostic yield across patient groups and predictive performance of clinical factors for genetic diagnosis.

(A) Bar plot showing the diagnostic yield (%) across different patient groups, considering index patients only. The segmental dystonia group includes patients with multifocal dystonia. (B) ROC curves illustrating the performance of individual predictors and a combined model for genetic diagnosis. The predictors include age at onset, generalized dystonia, additional features, positive family history, and a combined model incorporating all four predictors. Each curve shows the balance between sensitivity (true positive rate) and 1-specificity (false positive rate), with the Area Under the Curve (AUC) indicating predictive performance.