MATERIALS & METHODS
- Ethical considerations
All patients or their legal guardians gave informed consent to participate in this study. The research protocol was approved by the Ethics Committee of the Iran University of Medical Sciences (IR.IUMS.FMD.REC.1399.350 and IR.IUMS.FMD.REC. 1399.604). It was performed according to the ethical principles set out in the Declaration of Helsinki (1975/2008).
- Samples and data collection
This cross-sectional study, conducted at a tertiary movement disorders center between 2021 and 2024, included 34 patients with ChA from 24 unrelated Iranian families. Five patients (P2-1, P21, P22, P23, and P24) had been reported previously but were included because of their relevance to the present analyses [
9]. Patients were included based on compatible clinical features and the presence of
VPS13A variants identified through genetic testing. Variants were classified using the American College of Medical Genetics and Genomics (ACMG) criteria [
10], and cases with uncertain or conflicting variants were retained when clinical evidence strongly supported a ChA diagnosis and no alternative genetic cause was found. However, these cases were interpreted cautiously.
Demographic, clinical, and paraclinical data, including current age, age at onset (AAO), initial symptoms, and intervals from symptom onset to loss of ambulation or death, were collected through chart review and in-person assessments by a neurologist (M.R.). AAO was defined as the age at which the first clearly recognizable neurological or psychiatric symptoms attributable to ChA were identified, based on medical documentation and patient/caregiver reports. For patients with a positive family history who underwent presymptomatic genetic testing, AAO was determined retrospectively according to the onset of the first clinical manifestation rather than the time of genetic diagnosis. Motor manifestations were documented through neurological evaluation. For patients unable to attend visits because of an advanced disease, the most recent available medical records were used. Neuropsychiatric and behavioral features were ascertained through clinical evaluation during study visits, medical record review, and patient/caregiver report; these variables represented the lifetime documented history rather than cross-sectional symptoms at enrollment.
Patients were categorized as surviving ambulatory, surviving non-ambulatory, or deceased at the time of last follow-up. Nonambulatory status was defined as wheelchair dependence indoors and outdoors, or only very limited assisted indoor mobility with wheelchair dependence outdoors. Ambulatory status was defined as the ability to walk independently or with assistance/a walker sufficient for daily living activities. These cross-sectional categories were used solely for descriptive group comparisons and were distinct from the event definitions used in time-to-event analyses. Demographic, motor, behavioral, and other neurological characteristics were compared across these three groups.
- Genetic analysis
Whole-exome sequencing (WES) was performed on the proband’s DNA. Sequence data were analyzed and filtered as described previously (
Supplementary Material) [
11,
12]. Pathogenic variants in
VPS13A were investigated regardless of allele frequency. Copy number variations (CNVs) in neurodegenerative genes were identified using GATK tools [
11]. Candidate
VPS13A variants were validated by polymerase chain reaction (PCR) and Sanger sequencing, with co-segregation analysis in family members [
12]. Identified CNVs were further assessed using the Integrative Genomics Viewer (IGV) (
Figure 1).
Due to limited patient accessibility, reverse transcription PCR could not be performed to assess synonymous and splicing variants; therefore, NNSplice, SpliceAI, and Human Splicing Finder were used to predict potential splicing effects.
- Statistical analysis
Data are presented as median and interquartile ranges (IQR) for quantitative variables and as frequencies (percentages) for qualitative variables. Correlations between normally distributed continuous variables were evaluated using Pearson’s correlation coefficient, whereas the point-biserial correlation coefficient was used to assess relationships between binary and continuous variables. Categorical variables were analyzed using Pearson’s chi-square test or Fisher’s exact test, as appropriate, depending on expected cell counts. To account for the multiple comparisons conducted across numerous clinical variables, the Benjamini–Hochberg false discovery rate (FDR) correction was applied (q=0.05).
Survival and ambulation probabilities were estimated using the Kaplan–Meier method, defining the interval from disease onset to wheelchair dependence for ambulation and from disease onset to death for survival. Dates of disease onset, loss of ambulation, and death were retrospectively reconstructed from medical documentation and patient/caregiver reports. Dates of last follow-up were determined from the most recent documented evidence. Patients who remained alive at the end of follow-up were censored. For the ambulatory analysis, patients who remained ambulatory at the most recent clinical assessment were censored at that time. All patients with a documented transition to non-ambulatory status were counted as events irrespective of subsequent survival status. Deceased patients without documented loss of ambulation were censored at their last clinical evaluation.
All statistical analyses were conducted using IBM SPSS Statistics version 26 (IBM Corp.), with p<0.05 considered statistically significant.
RESULTS
- Demographic, clinical, and paraclinical findings
This study enrolled 34 patients clinically and paraclinically compatible with ChA. Eleven (32.4%) of the patients were female. Of these patients, 21 (61.8%) were born to consanguineous parents, and 21 (61.8%) had one or more siblings diagnosed with the same condition. The most prevalent presenting symptom was seizure (
n=12, 35.3%), followed by chorea, including limb and/or orofacial chorea (
n=10, 29.4%), behavioral disorder (
n=8, 23.5%), and involuntary vocalization (
n=4, 11.8%). The median current age, disease duration, and AAO for the 34 patients were 38.5 years (IQR, 34–43 years), 8.0 years (IQR, 4–12.25 years), and 29.0 years (IQR, 27–33.25 years), respectively.
Table 1 presents demographic data and detailed physical examination findings for the surviving ambulatory, surviving non-ambulatory, and deceased groups.
Of the 34 patients, 9 (26.5%) experienced loss of ambulation after a median of 10 years (range, 5–16 years) from disease onset, including one patient who became non-ambulatory three years before death. Detailed data for non-ambulatory patients are provided in
Supplementary Table 1. The Kaplan–Meier model estimated the probability of remaining ambulatory to be 87.4% (standard error [SE]=0.068) at 6 years, decreasing to 38.0% (SE=0.155) at 16 years after disease onset (
Figure 2A).
Of the 34 patients, 7 (20.6%) died after a median disease duration of 10 years (range, 3–17 years). Among the deceased patients, two died by suicide, while 4 (11.8%) of all patients attempted suicide; two patients died of sepsis, one died of status epilepticus, one died in a motor vehicle accident, and one died of an unknown cause. Detailed data for deceased patients are provided in
Supplementary Table 2. The Kaplan–Meier model estimated the survival probability to be 88.4% (SE=0.064) at 7 years, decreasing to 56.3% (SE=0.152) at 17 years after disease onset (
Figure 2B).
Fisher’s exact test indicated that patients with rigidity were more likely to be non-ambulatory (75.0% vs. 20.0%; p=0.048), and all non-ambulatory patients experienced falls (p=0.034). Only patients with insomnia reported suicide attempts (p=0.039). Low self-confidence was consistently associated with sad mood (p=0.011) and suicidal ideation (p=0.005); all affected patients exhibited sad mood, and all those with suicidal ideation had low self-confidence. Chi-square analysis further demonstrated a relationship between low self-confidence and irritability (χ2[1, n= 34]=5.384, p=0.020); 71.4% of patients with irritability showed low self-confidence compared with 30.8% of those without irritability.
Compulsions co-occurred with obsessions (100% of patients;
p<0.001) and were associated with a higher prevalence of suicidal ideation (46.2% vs. 9.5%;
p=0.033), while insomnia was more frequent in patients without compulsions (61.9% vs. 23.1%; χ
2[1, n=34]=4.859,
p=0.028). Involuntary tongue-lip biting (
Figure 3) was linked to drooling (75.0% vs. 38.9%; χ
2[1, n=34]=4.480,
p=0.034) and aggressive behavior (χ
2[1, n=34]=5.812,
p=0.016); 75.0% of patients with aggressive behavior exhibited tongue-lip biting compared with 31.8% of those without aggressive behavior.
Pearson’s correlation analysis demonstrated that disease duration was positively associated with ambulatory duration (r=0.907, p<0.001). Point-biserial correlations revealed that longer disease duration was related to non-ambulatory status (rpb=0.506, p=0.002), low self-confidence (rpb=0.390, p=0.023), falling (rpb=0.427, p=0.012), feeding dystonia (rpb=0.523, p=0.002), and dysarthria (rpb=0.479, p=0.004). Older age was positively associated with non-ambulatory status (rpb=0.525, p=0.001), falling (rpb=0.432, p=0.011), and dysarthria (rpb=0.357, p=0.038), while lower extremity chorea was negatively correlated with age (rpb=-0.343, p=0.047). Age at disease onset was inversely correlated with irritability (rpb=-0.444, p=0.009), orofacial chorea (rpb=-0.396, p=0.020), limb chorea (rpb=-0.417, p=0.014), and involuntary vocalization (rpb=-0.378, p=0.028).
After FDR correction for multiple point-biserial correlations and group comparisons, only the associations between age and non-ambulatory status, disease duration and non-ambulatory status, disease duration and feeding dystonia, and compulsions-obsessions retained statistical significance.
These symptoms are illustrated in
Supplementary Videos 1-
6. Brain MRI was performed in all patients and revealed atrophy of the caudate nuclei and putamen (
Figure 4).
Two individuals (P2-1 and P3) with advanced ChA underwent GPi-DBS for severe choreiform movements; both showed partial improvement.
- Genetic findings
A total of 21 variants were identified in 34 patients, including 12 novel variants (
Table 2), and all variants co-segregated with disease status in the families. The variants were distributed throughout
VPS13A rather than being confined to specific exons. According to ACMG criteria, eight were pathogenic, 11 likely pathogenic, and one was of uncertain significance (variant of uncertain significance [VUS]) (
Table 2), indicating variable levels of diagnostic certainty across cases. A challenging variant, c.3083A>G;p.Glu1028Gly, predicted to be “benign/ likely benign,” was found in patients P9-1, P9-2, and P14 in the compound heterozygous and homozygous states, respectively. Its frequency was 0.02246, 0.00420, 0.00180, and 0.000227 in Iranome, gnomAD4, the 1000 Genomes Project, and the ESP6500AA databases, respectively. In Iranome (1,202 individuals) and 200 in-house healthy WES samples, only one homozygote individual with c.3083A>G was identified, who was 30 years old and asymptomatic. The variant was also reported in six homozygote individuals in gnomAD4. It had a combined annotation-dependent depletion score of 21.8 and was predicted as possibly damaging by PolyPhen-2 and damaging by PROVEAN and FATHMM-MKL. Although the pathogenicity of this variant is conflicting, it was considered a possible contributory variant as it was identified in two families (P9 and P14) without other detectable
VPS13A single-nucleotide variants (SNVs) or CNVs. However, its pathogenicity remains unconfirmed, and these cases remain of uncertain molecular etiology.
A disease-causing variant was not identified through routine SNV analysis of WES data in P8. However, CNV analysis using IGV and Binary Alignment Map (BAM) files revealed a large homozygous deletion spanning exons 8 and 9 of
VPS13A (
Figure 1), which was not present in other ChA patients or in 200 controls.
The synonymous variant c.7806G>A, identified in P7, was predicted to affect splicing by NNSPLICE, Human Splicing Finder, and SpliceAI (
https://spliceailookup.broadinstitute.org/), and has been previously reported [
13]. This variant alters the 3' nucleotide of exon 55 (CCGgtaata to CCAgtaata), disrupting the splice-donor consensus despite both codons encoding proline. Its absence in more than 1,400 Iranian individuals supported its classification as disease-causing in this case.
DISCUSSION
Ambulation progressively deteriorates in patients with ChA because of chorea, dystonia, parkinsonism, and characteristic gait disturbances, such as the rubber man gait and stutter-step, as previously described [
14,
15]. These unpredictable movements increase the risk of falls. Falls were moderately to strongly correlated with disease duration, suggesting that falls become more frequent as the disease progresses. Consistent with previous reports of gait disturbances and postural instability [
2,
16-
18], the high frequency of falls observed in this study highlights the need to assess patients for injuries resulting from falls. Falls may also increase patients’ dependency and subsequent caregiver burden; therefore, walking aids and physiotherapy should be considered [
19]. Falls were also associated with dysarthria and feeding dystonia (
Supplementary Videos 1 and
2), reflecting a more generalized stage of motor impairment in affected patients.
The estimated median time to loss of ambulation was 16 years from disease onset, indicating that most patients maintain independent mobility for more than a decade. The probability of preserving ambulation was 87.4% six years after disease onset, which decreased to 38.0% at 16 years; these findings indicate that loss of motor function is gradual, highlighting the importance of early interventions to prolong functional independence. Patients with longer disease duration had more opportunity to progress through disease stages, resulting in a higher likelihood of reaching non-ambulatory status. Furthermore, the association of disease duration and falling with non-ambulatory status suggests an increased risk of loss of ambulation with disease progression and gait deterioration. Although previous reports indicate that some patients become wheelchair-dependent by the end of their third decade of life [
20], most patients in our study lost ambulation in their thirties.
Survival in ChA is highly variable and is influenced by the severity of motor and behavioral symptoms and the occurrence of complications, such as aspiration pneumonia. The mean survival duration was estimated at 16.7 years after disease onset. Similarly, Huang et al. [
6] reported a ChA disease duration of 12–20 years and a maximum survival of 40 years. Survival probability declined from 88.4% at 7 years to 56.3% at 17 years after disease onset, reflecting the progressive neurodegenerative nature of ChA and increasing mortality over time. Walker et al. [
8] reported a mean post-diagnosis survival of 11 years and noted shorter survival among patients diagnosed at very young or very old ages. This association was not observed in our cohort, likely because no patients had very early or very late disease onset.
The mean AAO was approximately 30 years, in line with previous reports [
8,
20]. In cases where ChA initially presents with cognitive or behavioral symptoms, the diagnosis may initially be mistaken for other disorders; even motor impairments may be incorrectly attributed to the side effects of the prescribed psychiatric medications [
2]. Thus, determining AAO in ChA is often challenging. Reported ages at onset probably correspond either to a significant event, such as a seizure, or to the point at which symptoms prompted medical attention [
8]. In our cohort, approximately 50% of patients without a family history of ChA initially presented with seizures. In cases with affected relatives, AAO may also have been influenced by the timing of genetic testing [
8].
ChA is generally regarded as a poor-prognosis and ultimately fatal disease in which sudden death or other complications of progressive neurodegeneration occur [
2,
6]. Walker et al. [
8] reviewed 52 patients with ChA; among the reported mortality causes, aspiration pneumonia was the most frequent (22%), followed by sudden unexplained death (19%), cardiac issues (16%), suicide-related death (13%), seizure (12%), asphyxiation/aspiration (9%), and others (9%). Muscle weakness, which has been reported in roughly 50% of ChA patients [
21], can increase the risk of all-cause mortality and reduce life expectancy [
22]. The reasons for death in this study were sepsis (
n=2), suicide (
n=2), seizure (
n=1), motor vehicle collision (
n=1), and one case of sudden death of unknown origin. Understanding the causes of death may help guide physicians toward more effective management and addressing potentially complicating symptoms that worsen patient outcomes.
Patients with feeding dystonia often use a head extension maneuver to propel food into the oropharynx, which increases the risk of aspiration pneumonia [
8,
23]. Gastrostomy placement is considered beneficial for preventing aspiration and nutritional deficiencies, which can result in weight loss [
2]. There are also case reports suggesting that severe oromandibular dystonia can be successfully improved by botulinum toxin application [
2]. Minimizing the risk of aspiration may play a crucial role in extending patients’ life expectancy.
Tongue-lip biting (
Figure 3) was observed in 31.6%, 62.5%, and 71.4% of the surviving ambulatory, surviving non-ambulatory, and deceased groups, respectively. It can result in oral cavity ulcers. Such ulcers may lead to severe complications, as a case of sepsis and intensive care unit admission was reported by Garrido-Fernández et al. [
24] following ulcers caused by tongue-lip-biting.
Psychiatric abnormalities, including anxiety, depression, bipolar disorder, and obsessive-compulsive disorder (OCD), have been reported in approximately half of the ChA patients [
6,
25]. We also observed a high prevalence of anxiety, depression, and obsessive symptoms in our patients, whereas delusion and hallucinations were uncommon. Addressing these symptoms is important in enhancing patients’ quality of life [
26,
27]. Psychiatric symptoms, with or without cognitive decline [
28], often precede movement disorders in ChA [
29]. These symptoms may present early, more than a decade before other neurological symptoms [
30]. Moreover, psychiatric symptoms may be the early manifestations of basal ganglia neurodegeneration and may contribute more substantially to disability than motor symptoms [
31]. Several significant associations among neuropsychiatric features were identified, underscoring their clinical significance and the need for careful evaluation.
Severe psychiatric symptoms may prompt consideration of suicide risk. Suicide accounted for approximately 10% of deaths in ChA patients [
8]. Suicidal behavior may represent a severe form of self-mutilation, which is thought to be a compulsive behavioral pattern [
8,
32]. In our cohort, suicidal ideation was reported in 8 of 34 patients (23.5%), and two patients with a history of depression and prior suicide attempts died by suicide in their thirties.
Despite the slow, progressive nature of ChA, sudden death can occur, potentially due to seizures or suicide [
2,
8], as one of our patients who had initially presented with seizures eventually died of status epilepticus. Seizure was the most common presenting symptom (35.3%). Overall, the prevalence of seizures was 57.9%, 50.0%, and 71.4% in the surviving ambulatory, surviving non-ambulatory, and deceased groups, respectively. Seizures are a distinguishing feature of ChA from most other choreatic disorders [
23] and may precede motor manifestations by over a decade [
33-
35]. Given the relatively high prevalence of seizures in ChA and the associated risk of sudden death, appropriate management is critically important.
DBS has been shown to be beneficial in the treatment of severe OCD, dystonia, and Parkinson’s disease. He et al. [
7] systematically reviewed the clinical outcomes of GPi-DBS in 27 patients with ChA. Their analysis showed that DBS can alleviate disabling motor symptoms, including gait disturbance and postural instability. The total motor score (TMS) of the Unified Huntington’s Disease Rating Scale (UHDRS) was significantly improved at both short- and long-term follow-ups. Although linear regression did not identify any reliable predictors of GPi-DBS efficacy, the authors suggested that patients predominantly presenting with oromandibular dystonia may benefit most from this intervention [
7]. Another study involving 15 patients with ChA who underwent GPi-DBS revealed significant improvements in both TMS and functional assessment scores at early and late postoperative follow-ups. In contrast to parkinsonian symptoms, dystonia and chorea showed marked improvement. Preoperative TMS severity was identified as a predictor of motor improvement in both postoperative phases [
36]. Liu et al. [
37] evaluated the UHDRS motor scores of six ChA patients before and following GPi-DBS. Compared with the preoperative assessments, both the mean TMS and the chorea subscore improved at follow-up. However, adverse effects were also reported. One patient experienced a generalized tonic-clonic seizure the day after surgery, and another developed dysarthria six months later.
In the present study, two patients underwent GPi-DBS. Both patients experienced partial improvement in motor symptoms (chorea improved by 40%–60% based on UHDRS). The first case, approximately 2.5 years after DBS surgery, had a spinal fracture and spinal cord compression following a seizure episode, leading to quadriplegia. Despite undergoing spinal surgery, the patient eventually died of severe sepsis. The second patient, who presented with severe chorea, tongue-lip biting, and seizures, underwent GPi-DBS 12 years after symptom onset. Although his movement disorders improved postoperatively, he developed status epilepticus and cardiorespiratory arrest approximately two years later, resulting in his death. Both patients had received appropriate antiepileptic therapy but ultimately died due to seizures or their complications. While both patients experienced motor improvement after GPi-DBS, their long-term outcomes remained poor, highlighting the variable and limited efficacy of GPi-DBS in advanced ChA, particularly in the presence of seizure comorbidity. Overall, further studies are needed to clarify the predictors of treatment efficacy and optimize candidate selection.
We also investigated genetic characteristics and clinical heterogeneity in ChA. Considerable clinical heterogeneity was observed, even among patients carrying the same variant. In family P9, for example, the AAO was 14 years in P9-1 and 29 years in P9-2; moreover, P9-1 exhibited truncal dystonia, tongue-lip biting, and seizures, whereas P9-2 had not yet developed these manifestations by age 36. These findings suggest that genetic, epigenetic, and environmental factors contribute to phenotypic variability [
38]. Similar variability has been reported in a ChA mouse model with a
VPS13A exon 60–61 deletion, in which differences in disease onset and phenotype were attributed to genetic background and modifier genes [
39-
41].
Approximately 26% of
VPS13A variants affect splicing, possibly due to the large number of splice sites in the gene (
n=143), which create numerous potential mutational sites [
13]. In this study, variants potentially affecting
VPS13A mRNA splicing included four pathogenic splice-site variants, one likely pathogenic large deletion, and one likely pathogenic synonymous variant. Although in-silico analyses predicted disrupted splicing, the precise transcript-level effects could not be determined because RNA samples from the patients were unavailable.
Among the identified variants, c.3083A>G;p.Glu1028Gly was particularly challenging because conflicting evidence exists regarding its pathogenicity. Although it was identified in affected individuals from two unrelated families without alternative VPS13A candidate variants, its relatively high population frequency and presence in homozygous individuals in population databases argue against a definitive classification as pathogenic. Therefore, this variant was cautiously considered a possible phenotype-contributing or low-penetrance variant rather than a confirmed disease-causing variant. Further functional and genetic studies are required to clarify its biological significance. Consequently, the molecular basis of disease in these families remains unresolved.
WES identified pathogenic or likely pathogenic VPS13A SNVs and CNVs in 21 of 24 families (diagnostic yield, 87.5%). A definitive genetic diagnosis was not established in three families: P5, which carried one likely pathogenic variant and one VUS (c.644A>G); P9, which carried one pathogenic variant and one likely benign variant (c.3083A>G); and P14, which harbored a homozygous likely benign variant (c.3083A>G).
Several identified variants were novel, which may contribute to a better understanding of disease mechanisms and the development of targeted therapies for this rare and debilitating neurodegenerative disorder.
This study has several limitations. Although ChA is more frequently reported in Iran, possibly because of the high prevalence of consanguineous marriages, it remains a rare disorder, resulting in a small sample size that limits statistical power and generalizability. Furthermore, only a limited number of findings remained significant after correction for multiple testing, indicating that nominal associations (uncorrected p<0.05) should be considered exploratory and require validation in larger cohorts. The UHDRS could not be administered to some patients because of advanced disease or being bedridden. In addition, severe neuropsychiatric and behavioral manifestations interfered with both treatment adherence and clinical assessment, potentially resulting in assessment bias. As neuropsychiatric data were derived from clinical evaluations, medical records, and patient/caregiver reports rather than standardized time-point assessments, correlation analyses must be interpreted cautiously. Moreover, some behavioral features, including low self-confidence, were based on clinician judgment rather than validated psychiatric scales, thereby limiting reproducibility. Finally, RNA samples were unavailable to determine the transcript level effects of splice variants, and the lack of functional studies means that the pathogenicity of certain variants, particularly c.3083A>G, requires further confirmation.
In conclusion, ambulatory impairment may be associated with survival and may serve as a marker of disease progression. Future longitudinal studies are warranted to clarify its relationship with survival and evaluate ambulation as a treatment outcome measure. Our findings also underscore the importance of managing seizures and psychiatric manifestations, which may substantially affect survival. Although disease-causing variants were identified in 21 families, a definitive genetic diagnosis was not established in three families, emphasizing the need for larger cohorts and multicenter studies to facilitate the interpretation and reclassification of uncertain variants.