Motor, Extrapyramidal, and Cognitive Involvement in RFC1 Disease Authors: Sara Massucco, Mehrnaz Hamedani, Marta Ponzano, Riccardo Manca, Edoardo Roveta, Anna Pichiecchio, Carlo Asteggiano, Federico Zaottini, Maria João Malaquias, Riccardo Currò, Matteo Pardini, Angelo Schenone, Federico Massa, Andrea Cortese, Marina Grandis Published: 2026 ABSTRACT Background and ObjectivesBiallelic intronic repeat expansions in the replication factor C subunit 1 (RFC1) gene are a common cause of cerebellar ataxia, neuropathy, vestibular areflexia syndrome and other late-onset ataxias. Recent evidence suggests a broader phenotypic spectrum. This systematic review and meta-analysis evaluated the prevalence and features of motor neuron, extrapyramidal, and cognitive involvement in RFC1 disease.MethodsWe systematically searched PubMed, Scopus, and Web of Science for studies reporting motor neuron, extrapyramidal, or cognitive involvement in RFC1 disease from inception to March 2025. Eligible sources comprised research articles and case reports/series on genetically confirmed cases. Extracted data included publication details, study design, location, and individual participant data (IPD) covering demographics, signs, and symptoms. When needed, IPD were obtained from corresponding authors. Random-effects models were used to estimate pooled prevalences (95% CIs). Univariable mixed-effects logistic regression accounted for clustering within studies, with multivariable models including variables with p < 0.10 in univariate analyses.ResultsOf 729 articles, 37 were included in the systematic review and 36 (874 patients) in the proportion meta-analyses. For IPD meta-analysis, 30 cohorts provided data on 312 patients (mean age 66.91 ± 10.76 years; median disease duration 12 [interquartile range 7–18] years; 51.5% male). Upper and lower motor neuron signs were present in 18% (95% CI 6%–34%) and 11% (95% CI 2%–24%). Patients with the ACAGG pentanucleotide repeat expansion (biallelic or compound heterozygous with an AAGGG expansion) had higher odds of muscle atrophy (OR 17.90, p 0.001) and weakness (OR 11.69, p 0.001); men had an increased likelihood of muscle atrophy (OR 2.71, p = 0.038). The pooled prevalence of parkinsonism and cognitive impairment was 8% (95% CI 1%–18%) and 31% (95% CI 13%–53%), with prominent executive-attention deficits. Cognitive decline was more likely with longer disease duration (OR 1.04, p = 0.047) and in biallelic AAGGG carriers vs ACAGG (ACAGGexp: OR 0.13, p = 0.041).DiscussionMotor neuron, extrapyramidal, and cognitive involvement may extend the phenotypic spectrum of RFC1 disease. Still, their true prevalence remains uncertain due to heterogeneity across studies, with potential overestimation from publication bias favoring atypical cases and underestimation when subtle signs are not systematically investigated. Larger multicenter cohorts with standardized assessments are needed to clarify their clinical relevance. Introduction Biallelic intronic AAGGG pentanucleotide repeat expansions in the replication factor C subunit 1 (RFC1) gene are a major cause of cerebellar ataxia, neuropathy, and vestibular areflexia syndrome (CANVAS), as well as other late-onset ataxias.1,2 With an estimated allele frequency of 1 in 14, the prevalence of biallelic expansion is approximately 1 in 712, making RFC1-related disease the most common autosomal recessive ataxia.3 Additional pathogenic repeat motifs have been identified, and expansion size has been shown to influence age at onset, clinical phenotype, and disease severity.4 Onset typically occurs around age 50 years with balance impairment.5 Patients develop axonal sensory neuronopathy, often with bilateral vestibular dysfunction, with cerebellar involvement emerging over time.5 More than 60% experience chronic spasmodic cough, sometimes decades before neurologic symptoms,5 while mild autonomic dysfunction affects about one-third.5 Since the identification of its genetic basis, reports have described features beyond the classical triad of ataxia, neuropathy, and vestibular dysfunction. These include upper motor neuron (UMN) and lower motor neuron (LMN) involvement,6-8 extrapyramidal signs,9-13 severe autonomic dysfunction,14-17 and cognitive impairment.18-20 However, most evidence comes from case reports or small series, leaving the prevalence of these atypical manifestations uncertain. This systematic review and meta-analysis aims to define the prevalence and features of the broader RFC1-related phenotype, with emphasis on motor neuron involvement, extrapyramidal signs, and cognitive and behavioral changes. Methods This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (eTable 1).21 Eligibility Criteria No publication date restrictions were applied. Inclusion criteria were defined using the Population, Intervention, Comparison, Outcome, and Study design (PICOS) framework. The Population (P) comprised individuals with genetically confirmed RFC1 disease, regardless of age, sex, disease severity, or duration. The Intervention (I) was defined as reports of motor neuron, extrapyramidal, or cognitive involvement. Motor neuron signs comprised hyperreflexia, Babinski sign, spasticity, fasciculations, or electromyographic (EMG) evidence of denervation. Extrapyramidal features included parkinsonism, dystonia, chorea, dyskinesias, or abnormal dopamine transporter imaging. Cognitive involvement was defined as mild cognitive impairment or dementia. The Comparison (C) was not required, as the objective was to characterize atypical features within the RFC1 population. The primary Outcome (O) was the prevalence of motor neuron, extrapyramidal, and cognitive manifestations in RFC1 disease. Secondary outcomes included: (1) characterization of clinical features in patients with these atypical signs and (2) analysis of potential modifiers such as sex, age, age at onset, disease duration, and repeat motif in the RFC1 gene. Outcome measures consisted of clinical findings, rating scales, neurophysiology, neuropsychology, and imaging. Eligible Study designs (S) encompassed original research, case reports/series, and cohort studies. Reviews, meta-analyses, editorials, conference abstracts, and nonhuman studies were excluded. Only English-language publications were considered. Search Strategy We conducted a comprehensive search of PubMed, Scopus, and Web of Science from database inception to March 29, 2024, with an update on March 17, 2025. The strategy incorporated keywords and Boolean operators aligned with the inclusion criteria (eMethods). Study Selection and Data Collection Study selection was performed systematically using the Rayyan software. Two authors (SM, MH) independently screened titles and abstracts, followed by full-text review of potentially eligible articles. Publication details and patient-level data (demographics, clinical signs, neuropsychological, neurophysiologic, and imaging findings) were extracted from the included studies. When necessary, corresponding authors were contacted for additional information (eAppendix 1). Both qualitative (case reports) and quantitative (clinical trials, cohort studies) studies were analyzed to provide a comprehensive overview of atypical RFC1 manifestations. For quality assessment, 2 authors (S.M., M.H.) independently applied the Joanna Briggs Institute Critical Appraisal Checklist (JBIC) for prevalence studies, case reports, and case series.22 The overall certainty of evidence was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.23 The study protocol was registered with PROSPERO before data extraction (CRD420251062794). Statistical Analysis We conducted a meta-analysis of proportions to estimate the prevalence of relevant symptoms and signs, calculating pooled prevalence with 95% CIs. Data were included from studies that reported prevalence rates or provided patient-level information enabling calculation. Case reports/series and studies assessing the prevalence of RFC1 expansions in Parkinson disease (PD), multiple system atrophy (MSA), or motor neuron disease (MND) were excluded from proportion meta-analyses and evaluated in sensitivity analyses. Meta-analyses were conducted using a random-effects model to account for heterogeneity across studies, with Freeman-Tukey double arcsine transformation and inverse variance weighting to stabilize variances and address extreme rates. Forest plots were generated, and heterogeneity was evaluated using the Q statistic and I2, with I2 > 50% considered substantial. As a secondary analysis, we conducted individual participant data (IPD) meta-analyses to examine the impact of patient characteristics on outcomes. A one-stage IPD meta-analysis was performed using mixed-effects logistic regression models accounting for the clustering of participants within each study. For each outcome, multivariable models included variables with p 0.10 in univariable models. Analyses were performed in Stata v.18 (StataCorp). Reporting followed the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) checklist (eTable 2).24 Role of the Funding Source The funder had no role in study design, data collection, analysis, interpretation, or manuscript preparation. Standard Protocol Approvals, Registrations, and Patient Consents This study used anonymized IPD from published studies; no new participants were enrolled. Informed consent was obtained in the original studies per institutional guidelines. Data Availability Data are available from the corresponding author upon reasonable request. Results Studies Included The search yielded 729 articles (Figure 1). After removing duplicates, 575 remained for title and abstract screening. Fifty-nine underwent full-text review, of which 31 met the inclusion criteria. A search update a year later identified 6 additional papers, for a total of 37 (Table 1; eAppendix 2). Of these, 19 research articles and 11 case reports/series reported on motor, extrapyramidal, or cognitive involvement in RFC1 disease. Four studies examined the prevalence of RFC1 expansions in PD, 2 in MSA, and 1 in MND. Since most publications were case reports/series or small cohorts, reporting bias is likely and could skew the prevalence estimates, with atypical cases more often reported and common presentations potentially underreported. Quality assessment using the JBIC and GRADE is presented in eTables 3–5. Given the small sample sizes and heterogeneity, overall study quality was rated low. Study Selection Flowchart Flow of studies through the screening process, from database search to title/abstract screening, full-text assessment, and final inclusion. Table 1Included StudiesFirst author yearrefCountryStudy designSample size—Eligible patients (N)1Liu 202539ChinaProspective cross-sectional multicenter national (prevalence of biallelic RFC1 repeat expansion in PD—retrospective collection of clinical data)22Abreu 202426PortugalCase series23Albajar 202442SpainProspective cross-sectional single center184Alvarez Jerez 202437United States (non-Finnish European ancestry)Prospective cross-sectional multicenter international (prevalence of biallelic RFC1 repeat expansion in PD—retrospective collection of clinical data)45Currò 20244International (white)Prospective multicenter international3926Dujardin 202420FranceProspective cross-sectional single center217Horowitz 202438FranceCase series68Izumi 20246JapanCase series39Li 202441ChinaProspective cross-sectional multicenter national (prevalence of biallelic RFC1 repeat expansion in MSA—retrospective collection of clinical data)110Schaub 202440GermanyProspective cross-sectional multicenter national (prevalence of biallelic RFC1 repeat expansion in MND—retrospective collection of clinical data)311Tsuchiya 202425JapanCase report112Azevedo 20239BrazilProspective cross-sectional single center1313Fonte 202313PortugalCase report214Malaquias 202327PortugalRetrospective multicenter national6715Ronco 202343International (white)Case series716Ylikotila 202311FinlandProspective multicenter national (prevalence of biallelic RFC1 repeat expansion in PD—both retrospective and prospective collection of clinical data)317Ando 202228JapanProspective multicenter national (retrospective collection of clinical data)1518Barghigiani 202229ItalyProspective single center919Colucci 202230ItalyCase series520da Silva Schmitt 202215BrazilProspective single center case-control design1621Herrmann 202232GermanyProspective single center422Huin 20228FranceProspective multicenter national3823Kermorvant 202233FranceCase report124Korpioja 202219FinlandRetrospective multicenter national925Kytövuori 202210FinlandProspective multicenter national (prevalence of biallelic RFC1 repeat expansion in PD with a case-control design)326Miyatake 202231JapanProspective multicenter national1627Record 202214United KingdomCase report128Reyes-Leiva 20227SpainCase report129Traschütz 202244GermanyCase report and prospective single center (Not included in proportion meta-analyses; duplicate cohort; see Traschütz 2021)7030Watanabe 202218JapanCase series731Matos 202134BrazilProspective multicenter national, case-control design2232Sullivan 202116United KingdomProspective single center (prevalence of biallelic RFC1 repeat expansion in MSA—retrospective collection of clinical data)333Traschütz 202135EuropeProspective multicenter international, double cohort design7034Cortese 20205International (white)Prospective multicenter international10035da Silva Schmitt 202012BrazilCase report136Scriba 202036Asia-pacificProspective multicenter international—retrospective collection of clinical data337Wan 202017ChinaProspective single center5Abbreviations: MND = motor neuron disease; MSA = multiple system atrophy; PD = Parkinson disease. Included Studies Abbreviations: MND = motor neuron disease; MSA = multiple system atrophy; PD = Parkinson disease. Patients Included In total, 874 patients were included in the proportion meta-analyses (808 in primary analyses; 874 in sensitivity analyses including case reports/series and MND/PD/MSA cohorts).4-20,25-43 The IPD meta-analysis included 312 patients from 30 studies (eTable 6),6-14,16-20,25-28,30-33,35-41,44 51.5% males and 48.5% females, with a mean age of 66.9 ± 10.8 years, mean age at onset of 53.4 ± 11.3 years, and median disease duration of 12 years (interquartile range (IQR) 7–18). A positive family history was reported in 47.5% (105/221). Most patients (277/312, 88.8%) carried biallelic intronic AAGGG expansions, 21 (6.7%) had biallelic ACAGG expansions, and 10 (3.2%) carried AAGGGexp/ACAGGexp. Rare genotypes included AAGGGexp/AAAGG15AAGGGexp and AAGGGexp/AAAGGexp with >500 AAAGG repeats (1.3% combined). Motor Neuron Involvement Pooled prevalence was 18% (95% CI 6%–34%) for UMN signs and 11% (95% CI 2%–24%) for LMN signs (Figure 2; eTable 7). Among LMN features, fasciculations occurred in 15% (95% CI 1%–34%) and EMG evidence of denervation in 40% (95% CI 8%–76%). UMN features included unspecified pyramidal signs in 18% (95% CI 6%–34%), brisk reflexes in 21% (95% CI 10%–34%), spasticity in 2% (95% CI 0%–5%), and Babinski sign in 2% (95% CI 0%–7%). Pooled Prevalences of Upper (A) and Lower (B) Motor Neuron Signs in RFC1 Disease Forest plots showing the pooled prevalence estimates (with 95% CIs) of upper (A) and lower (B) motor neuron signs across included studies. Study weights are reported in the right hand column; the diamond indicates the overall pooled prevalence. I2 quantifies between-study heterogeneity. The pooled prevalences of other features suggestive of motor involvement, but not specifically attributable to UMN or LMN involvement, are summarized in eTable 8. Muscle atrophy was reported in 8% (95% CI 1%–21%) and weakness in 5% (95% CI 0%–16%). Painful muscle cramps occurred in 16% (95% CI 4%–31%) and typically preceded ataxia (mean onset 34.5 ± 11.9 years), most often affecting the lower limbs; trunk (n = 2/19) and diffuse (n = 2/19) involvement were rare. Electrophysiology showed reduced compound muscle action potential amplitude in 28% (95% CI 14%–45%). Spontaneous activity on needle EMG was reported in 39% (95% CI 0%–96%). Central motor conduction time was bilaterally prolonged in 2 of 5 patients who underwent motor evoked potential testing: one study reported abnormal recordings from the abductor digiti minimi and another from the tibialis anterior.32,35 The median serum creatine kinase was 146.5 IU/L (IQR 105.3–526.5; n = 12), and hyperCKemia occurred in 30% (95% CI 0%–93%). Respiratory function was normal in 3 assessed patients.6 Imaging revealed spinal cord atrophy in 32/74 cases.5,6,27,32,40,43 In most reports, the anatomic distribution of atrophy (anterior cord/anterior horns vs posterior columns) was not specified; T2 hyperintensity of the posterior columns was described in 5 patients,5,43 and 1 study reported a reduced anteroposterior spinal cord diameter without further localization.6 When segmental distribution was reported,5,6 atrophy predominantly involved the cervical cord (n = 20/23), with fewer cases affecting the cervicothoracic (2/23) or thoracic (1/23) regions. Among patients with spinal cord atrophy, reported motor features included muscle atrophy or weakness (n = 4), fasciculations (n = 4), cramps (n = 3), and pyramidal signs (n = 7), including limb spasticity in 4. Muscle biopsy showed neurogenic changes in 5/21 cases, including group atrophy and fiber-type grouping.6,36 Postmortem neuropathology (n = 3)7,8,31 showed heterogeneous findings: axonal swellings at LMN contact points,8 moderate LMN loss in the hypoglossal nucleus and thoracic anterior horn without corticospinal tract degeneration,7 and severe posterior column neuronal loss with anterior horn sparing.31 The ACAGG expansion, either in the biallelic state or combined with AAGGG, increased odds of muscle atrophy (OR 17.90; 95% CI 4.83–55.33; p 0.001), weakness (OR 11.69; 95% CI 2.98–45.90; p 0.001), and fasciculations (OR 18.22; 95% CI 1.49–222.43; p = 0.023; Table 2). Male sex was associated with muscle atrophy (OR 2.71; 95% CI 1.06–6.97; p = 0.038) and fasciculations (OR 3.25; 95% CI 1.13–9.34; p = 0.029). Longer disease duration increased the odds of fasciculations by 6% per year (OR 1.06; 95% CI 1.01–1.11; p = 0.030). Older age at onset and older age at assessment were each associated with lower odds of UMN signs (OR 0.95; 95% CI 0.92–0.98; p = 0.004 for each; Table 2), with a 5% decrease in odds per additional year. Table 2Association of Sex, Age, Age at Onset, Disease Duration, and Repeat Motif in the RFC1 Gene With the Odds of Observing Motor SignsMuscle atrophyUnivariableMultivariable N = 281 (25)N pt (papers)OR (95% CI)p ValueOR (95% CI)p ValueMale sex283 (25)2.71 (1.06–6.97)0.038*3.80 (1.35–10.68)0.011*Age292 (26)1.02 (0.98–1.06)0.333——Age at onset280 (23)1.00 (0.96–1.04)0.952——Disease duration272 (23)1.02 (0.98–1.07)0.356——ACAGGexp296 (26)17.90 (4.83–66.33)<0.001***38.99 (6.69–227.18)<0.001***Muscle weaknessUnivariableN pt (papers)OR (95% CI)p ValueMale sex285 (26)1.12 (0.40–3.13)0.834Age294 (27)1.02 (0.97–1.07)0.434Age at onset282 (24)0.99 (0.95–1.04)0.810Disease duration274 (24)1.04 (0.99–1.10)0.148ACAGGexp297 (27)11.69 (2.98–45.90)<0.001***FasciculationsUnivariableMultivariable N = 254 (22)N pt (papers)OR (95% CI)p ValueOR (95% CI)p ValueMale sex283 (25)3.25 (1.13–9.34)0.029*3.34 (1.05–10.63)0.041*Age292 (26)1.03 (0.98–1.08)0.291——Age at onset280 (23)0.98 (0.94–1.02)0.350——Disease duration272 (23)1.06 (1.01–1.11)0.030*1.06 (1.00–1.12)0.036*ACAGGexp296 (26)18.22 (1.49–222.43)0.023*29.04 (0.93–911.20)0.055Pyramidal signsUnivariableMultivariable N = 276 (25)N pt (papers)OR (95% CI)p ValueOR (95% CI)p ValueMale sex283 (25)1.92 (0.91–4.03)0.0861.92 (0.90–4.13)0.093Age292 (26)0.95 (0.92–0.98)0.004**0.95 (0.92–0.99)0.016*Age at onset280 (23)0.95 (0.92–0.98)0.004**——Disease duration272 (23)0.99 (0.96–1.03)0.794——ACAGGexp296 (26)0.73 (0.17–3.04)0.662——SpasticityUnivariableN pt (papers)OR (95% CI)p ValueMale sex283 (25)1.83 (0.45–7.48)0.399Age292 (26)0.99 (0.93–1.05)0.674Age at onset280 (23)0.99 (0.93–1.05)0.720Disease duration272 (23)0.99 (0.91–1.08)0.834ACAGGexp296 (26)1.11 (0.13–9.21)0.921Babinski signUnivariableN pt (papers)OR (95% CI)p ValueMale sex280 (24)3.92 (0.70–21.93)0.120Age289 (25)0.99 (0.92–1.06)0.792Age at onset277 (22)0.96 (0.89–1.03)0.294Disease duration269 (22)1.03 (0.95–1.12)0.483ACAGGexp293 (25)3.65 (0.45–29.59)0.226Abbreviations: N pt = number of patients (number of studies in parentheses); OR = odds ratio; --- = not included in the multivariable model due to nonsignificance in univariable analyses or collinearity; *p < 0.05; **p < 0.01; ***p < 0.001.The ACAGG repeat expansion refers to both homozygous carriers and compound heterozygotes with AAGGG; odds ratios are calculated relative to the most common AAGGG expansion. Univariable and multivariable mixed-effect random-intercept logistic regression models were used. Association of Sex, Age, Age at Onset, Disease Duration, and Repeat Motif in the RFC1 Gene With the Odds of Observing Motor Signs Abbreviations: N pt = number of patients (number of studies in parentheses); OR = odds ratio; --- = not included in the multivariable model due to nonsignificance in univariable analyses or collinearity; *p < 0.05; **p < 0.01; ***p < 0.001. The ACAGG repeat expansion refers to both homozygous carriers and compound heterozygotes with AAGGG; odds ratios are calculated relative to the most common AAGGG expansion. Univariable and multivariable mixed-effect random-intercept logistic regression models were used. Extrapyramidal Involvement The pooled prevalence of parkinsonism was 8% (95% CI 1%–18%; Figure 3; eTable 9). Parkinsonism was reported in 54 patients, most often asymmetric (78.3%, 18/23). The median Unified Parkinson's Disease Rating Scale part III (UPDRS-III) score was 17.5 (IQR 12.5–36.0; n = 16). Symptom prevalence included resting tremor in 6% (95% CI 0%–17%), bradykinesia in 17% (95% CI 8%–27%), rigidity in 14% (95% CI 4%–27%), hypomimia in 9% (95% CI 0%–37%), and REM sleep behavior disorder (RBD) in 5% (95% CI 0%–12%). Hyposmia was reported in 3 cases.18,37,39 Of 21 patients treated with levodopa, 16 showed some degree of improvement in bradykinesia or rigidity, with a pooled estimate of 92% (95% CI 33%–100%; eTable 10). Among 7 cases with details,9,30,38,39 4 showed mild and 3 marked levodopa responses. Five patients developed on-off fluctuations,8,11,38,39 and 2 improved after deep brain stimulation (DBS).11,39 Pooled Prevalences of Parkinsonism in RFC1 Disease Forest plot showing the pooled prevalence estimates (with 95% CIs) of parkinsonism across included studies. Study weights are reported in the right hand column; the diamond indicates the overall pooled prevalence. I2 quantifies between-study heterogeneity. Dopamine transporter imaging showed reduced presynaptic uptake in 32/38 patients, including 9 without clinical parkinsonism, with a pooled prevalence of 65% (95% CI 9%–100%). Decreased uptake was typically bilateral but asymmetric and more pronounced in the putamen. Myocardial 123I-metaiodobenzylguanidine ([123I]MIBG) scintigraphy demonstrated postganglionic sympathetic denervation in 80% (single study),31 with 1 additional case report.25 None of these patients had parkinsonism; all showed sensory neuropathy and ataxia, 2 had cognitive impairment, and 3 dysautonomia. Two underwent dopamine transporter imaging, which was normal in one and reduced in the other. Structural imaging demonstrated basal ganglia atrophy (eAppendix 3).34 A pseudo-eye-of-the-tiger sign was seen in 2 patients,26 defined as a symmetric T2 hyperintensity in the globus pallidus surrounded by a linear isointense rim, without the iron-related peripheral T2 hypointensity typical of pantothenate kinase-associated neurodegeneration. Three patients with ataxia showed a hot cross bun sign, i.e., a cruciform T2 hyperintensity in the pons, thought to reflect pontocerebellar fiber degeneration.16,17,28 Two had dysautonomia and were misdiagnosed with MSA (one also had RBD and cognitive impairment17 and the other asymmetric parkinsonism unresponsive to levodopa16). The third had sensory neuropathy, myoclonus, pyramidal signs, muscle atrophy, weakness, and fasciculations without dysautonomia.28 The α-synuclein seed amplification assay was negative in 2 of 3 RFC1 cases,37 and phosphorylated α-synuclein immunofluorescence on skin biopsy was negative in both tested patients.39 Univariate models showed no association between parkinsonism and age, sex, age at onset, disease duration, or repeat motif in the RFC1 gene (Table 3). Table 3Association of Sex, Age, Age at Onset, Disease Duration, and Repeat Motif in the RFC1 Gene With the Odds of Observing Parkinsonism, Cognitive Impairment, and Psychiatric SymptomsParkinsonismUnivariableN pt (papers)OR (95% CI)p ValueMale sex292 (28)1.38 (0.59–3.27)0.458Age302 (29)0.99 (0.95–1.04)0.785Age at onset290 (26)0.97 (0.94–1.01)0.140Disease duration282 (26)1.03 (0.98–1.08)0.202ACAGGexp305 (29)1.20 (0.09–16.08)0.888Cognitive impairmentUnivariableMultivariable N = 274 (26)N pt (papers)OR (95% CI)p ValueOR (95% CI)p ValueMale sex288 (28)1.02 (0.53–1.97)0.950——Age298 (29)1.01 (0.97–1.04)0.681——Age at onset286 (26)0.99 (0.96–1.02)0.388——Disease duration278 (26)1.04 (1.00–1.08)0.0681.04 (1.00–1.08)0.047*ACAGGexp301 (29)0.20 (0.04–0.98)0.048*0.13 (0.02–0.92)0.041*Psychiatric symptomsUnivariableMultivariable N = 278 (26)N pt (papers)OR (95% CI)p ValueOR (95% CI)p ValueMale sex284 (26)0.94 (0.33–2.65)0.905——Age294 (27)0.97 (0.92–1.02)0.261——Age at onset282 (24)0.93 (0.89–0.98)0.005**1.00 (0.96–1.04)0.955Disease duration274 (24)1.07 (1.01–1.13)0.015*1.04 (0.99–1.09)0.135ACAGGexp298 (27)0.25 (0.01–8.50)0.440——Abbreviations: N pt = number of patients (number of studies in parentheses); OR = odds ratio; --- = not included in the multivariable model due to nonsignificance in univariable analyses or collinearity; *p < 0.05; **p < 0.01.The ACAGG repeat expansion refers to both homozygous carriers and compound heterozygotes with AAGGG; odds ratios are calculated relative to the most common AAGGG expansion. Univariable and multivariable mixed-effect random-intercept logistic regression models were used. Association of Sex, Age, Age at Onset, Disease Duration, and Repeat Motif in the RFC1 Gene With the Odds of Observing Parkinsonism, Cognitive Impairment, and Psychiatric Symptoms Abbreviations: N pt = number of patients (number of studies in parentheses); OR = odds ratio; --- = not included in the multivariable model due to nonsignificance in univariable analyses or collinearity; *p < 0.05; **p < 0.01. The ACAGG repeat expansion refers to both homozygous carriers and compound heterozygotes with AAGGG; odds ratios are calculated relative to the most common AAGGG expansion. Univariable and multivariable mixed-effect random-intercept logistic regression models were used. Autonomic dysfunction was common (38%; 95% CI 16%–62%), and some patients were misdiagnosed as MSA-P or MSA-C.5,16,17,41 Vertical gaze palsy was reported in 6 patients27,30,35 and retropulsion in 27% (single study).35 Hyperkinetic movement disorders were less frequent (3%; 95% CI 0%–10%; eTable 9). The estimated prevalence of chorea was 2% (95% CI 0%–9%), dystonia 5% (95% CI 2%–9%), and dyskinesias 4% (95% CI 1%–8%; eTable 10). Dystonia most often affected the hands and feet (n = 6) but generalized (n = 2), orofacial (n = 2), trunk (n = 1), and vocal (n = 2) involvement also occurred. When specified (n = 3),27 dyskinesias were oro-mandibular. Cognitive and Psychiatric Features Cognitive impairment was reported in 100 patients, with a pooled prevalence of 31% (95% CI 13%–53%; Figure 4, eTable 11). The median education among affected patients was 11.5 years (IQR 11–13; n = 22). In multivariate analysis (Table 3), longer disease duration increased the odds of cognitive impairment by 4% per year (OR 1.04; 95% CI 1.00–1.08; p = 0.047), while the ACAGG motif was associated with lower odds compared with biallelic AAGGG (OR 0.13; 95% CI 0.02–0.92; p = 0.041). Pooled Prevalence of Cognitive Impairment in RFC1 Disease Forest plot showing the pooled prevalence estimate (with 95% CI) of cognitive impairment across included studies. Study weights are reported in the right hand column; the diamond indicates the overall pooled prevalence. I2 quantifies between-study heterogeneity. Cognitive deficits were generally mild (58%, 21/36), most often affecting executive and attention/working memory domains. The median scores were 25 on the Mini-Mental State Examination (MMSE, range 0–30; IQR 24–29; n = 17), 26 on the Montreal Cognitive Assessment (MoCA, range 0–30; IQR 23–28; n = 34), and 14 on the Frontal Assessment Battery (range 0–18; IQR 11.45–16.35; n = 11). When detailed (n = 22),10,19,27,30,35,38 11 patients showed attentional/executive dysfunction, 5 had mental slowing, 3 an amnestic profile, and 3 aphasia (progressing to near mutism in 1). Ideomotor apraxia was noted in 2 cases. Among 39 systematically assessed patients,20,42 76.9% met criteria for definite cerebellar cognitive affective syndrome (CCAS) and 15.4% for probable CCAS; overall, 92.3% fulfilled criteria for either probable or definite CCAS. Magnetic resonance imaging (MRI) showed cerebral atrophy in 16% (95% CI 6%–28%), with heterogeneous regional distribution (eAppendix 3). Quantitative 3T MRI findings confirmed pronounced diffuse cerebellar atrophy and deep cerebral white matter involvement, while cortical damage appeared relatively limited.34 Functional imaging and CSF findings were reported in small numbers and are summarized in eAppendix 3; overall, they showed frequent cerebellar hypometabolism or hypoperfusion, occasional cortical involvement, and generally normal CSF profiles, although isolated elevations of neurofilament light chain and tau were observed.6,18,20,25,28,31,33,38,40,44 Psychiatric and behavioral symptoms were reported in 34 patients, with a pooled prevalence of 18% (95% CI 3%–38%; eTable 12). Depression (n = 15) and anxiety (n = 11, including 1 with panic attacks) were most frequent. Other features encompassed behavioral change, disinhibition, hypersexuality, or irritability (n = 5); apathy (n = 3); hallucinations (n = 3); alcohol abuse (n = 2); delusions (n = 2); pseudobulbar affect (n = 1); and mutism (n = 1). One Italian family presented with behavioral onset.30 Sensitivity analyses of pooled prevalence estimates are presented in eTable 13; eTable 14 summarizes the genotypes of patients included in proportion meta-analyses. Discussion Motor Neuron Involvement in RFC1 Disease RFC1 disease most often presents with the CANVAS phenotype, but motor neuron involvement has also been reported. This was first identified in Asia-Pacific patients carrying the ACAGG repeat motif31,36 and later reported in cohorts with biallelic AAGGG expansions.8,15,33 The preservation of tendon reflexes in many patients was initially attributed to sparing of large-caliber sensory fibers innervating muscle spindles.5 However, limb spasticity and Babinski signs observed in a subset of cases suggest pyramidal tract involvement. In this meta-analysis, about 1 in 5 patients showed UMN signs, with similar prevalences across RFC1 expansion motifs. Yet, pyramidal signs were often reported without clear definitions and may have simply reflected brisk tendon reflexes. Given the small number of patients with definite spasticity or Babinski signs and the relatively advanced age of most cohorts, UMN signs could reflect comorbid conditions common in older adults, such as cerebrovascular disease or degenerative spine pathology, complicating attribution solely to RFC1 pathology. However, the lower frequency of UMN signs with increasing age is not consistent with vascular comorbidity being the sole explanation. Moreover, brisk reflexes appear more frequent than expected in peripheral neuropathies, suggesting that this may represent a genuine feature of the RFC1 phenotype. The pooled prevalence of LMN involvement was 11%, although the true frequency may be underestimated given the limited availability of EMG data. Most studies did not specify the distribution of denervation findings, raising the possibility that some abnormalities could reflect common age-related conditions (e.g., degenerative spine disease with radiculopathy) rather than LMN involvement attributable to RFC1. Consistent with earlier reports,28,31,36 the ACAGG motif, whether biallelic or combined with AAGGG, was strongly associated with muscle atrophy, weakness, and fasciculations. Male sex also increased the odds of atrophy and fasciculations. Neuropathologic findings were heterogeneous. Some autopsies revealed anterior horn motor neuron loss,7,8 whereas others did not.31 Motor neuron loss in the hypoglossal nucleus and anterior horns without corticospinal tract degeneration has been described, suggesting that dysarthria and dysphagia may partly reflect LMN pathology.7 Swelling of synapses between UMNs and LMNs in the lumbar anterior horns has also been reported, and a synaptic dysfunction mechanism was proposed.8 Motor neuron involvement may occur in RFC1 disease, but biallelic RFC1 expansions appear to be uncommon in MND cohorts. In >1,000 sporadic amyotrophic lateral sclerosis (ALS) cases, no biallelic AAGGG expansions were detected (1 AACGGexp/AAAGGexp carrier),45 whereas in a German cohort of 107 MND patients, 3 (2 ALS, 1 primary lateral sclerosis) harbored biallelic AAGGG expansions.40 Extrapyramidal Involvement in RFC1 Disease In our meta-analysis, the pooled prevalence of parkinsonism in RFC1 disease was 8%. This rate is higher than the 0.2–0.3% in the general population and the 1–2% among individuals over 65 years46 and may suggest a genuine enrichment of nigrostriatal involvement in biallelic RFC1 expansion carriers. However, these figures derive from a limited number of studies and heterogeneous methodologies, thereby introducing potential bias that may inflate prevalence estimates. This concern is further supported by the GRADE assessment, which rated the overall quality of evidence for parkinsonism in RFC1-related disease as low for serious risk of selection bias (single-center referral cohorts), inconsistency and imprecision (high between-study heterogeneity in prevalence rates with wide CIs), as well as indirectness stemming from inclusion of PD-selected and MSA-selected populations alongside CANVAS cohorts. Moreover, the studies differ in terms of clinical definitions, driving prevalence estimates from 0-0.5% in large cohorts of biallelic RFC1 expansion carriers4,5 to 100% in small PD-selectced10,11,37,39 or MSA-selected series,16 highlighting referral biases. When all study types, including case reports/series and studies enriched for PD/MSA/MND, are pooled, the estimated prevalence rises to 34% (95% CI 16–54%; I2 ≈ 90%) vs 8% (I2 ≈ 87%) with restrictive inclusion, underscoring the influence of study populations. Individual motor signs showed similarly heterogeneous pooled rates (resting tremor 6%, bradykinesia 17%, rigidity 14%), reflecting varied ascertainment (e.g., bedside examinations with or without standardized UPDRS-III) and mixed tracers/readouts across centers. High levodopa responsiveness (92%; 95% CI 33–100%) and occasional DBS benefit support dopaminergic circuit involvement in at least a subset, consistent with a pure PD phenotype fulfilling Movement Disorder Society (MDS) criteria for PD in biallelic RFC1 expansion carriers.47 Similarly, cardiac [123I]MIBG scintigraphy frequently shows postganglionic denervation (≈80% of tested cases), including in patients without parkinsonism but with sensory neuropathy/ataxia, emphasizing a peripheral autonomic neuropathy distinct from classical synucleinopathies. Indeed, the multisystem nature of RFC1 disease (sensory neuronopathy, cerebellar degeneration, small-fiber autonomic failure), together, can produce MSA-like phenotypes in the absence of primary synucleinopathy, thereby complicating disease classification. Brain imaging also points to mixed, not unequivocal mechanisms. Presynaptic dopamine imaging was abnormal in approximately two-thirds (65%; 95% CI 9%–100%), typically with putaminal predominance and asymmetry. Nonetheless, subclinical presynaptic deficits on dopamine transporter imaging in patients without parkinsonism9 and normal dopamine transporter uptake in some patients with CANVAS and clinically diagnosed parkinsonism38 argue that basal ganglia denervation is the sole driver yet allow for the potential contribution of multisystem mechanisms beyond the nigrostriatal pathway, possibly involving the cerebellar-basal ganglia-brainstem network. Attribution of an accurate estimate of extrapyramidal impairment in RFC1 disease is also strongly age-confounded. Across CANVAS cohorts, typical onset falls in the 6th–7th decades, overlapping the window in which idiopathic parkinsonism and age-related motor slowing become common. In this age range, core RFC1 features can mimic or amplify parkinsonian signs (e.g., slowness, rigidity-like tone, gait hesitation, postural instability/retropulsion), inflating clinical estimates even without primary nigrostriatal disease. Notably, postural instability and retropulsion were frequent in large RFC1 series without formal PD diagnosis (49% and 27%, respectively, in Traschütz et al.,35 2021). By contrast, early onset PD-ascertained RFC1 carriers10,11,39 showed a pure PD phenotype according to MDS PD criteria,47 indicating that RFC1 can also underlie true nigrostriatal degeneration independent of late-life morbidity. Consistent with a nonuniform pathobiology, ancillary synuclein markers are frequently negative or variable, and olfactory testing is likewise inconsistent.18,37,39 Beyond parkinsonism, dystonia, limb chorea, and other hyperkinetic movements broaden the spectrum and are compatible with dysfunction within cerebellar-basal ganglia circuits secondary to cerebellar or brainstem pathology, or with direct basal ganglia involvement. Neuropathology—although limited—mirrors this heterogeneity. Huin et al.8 reported severe substantia nigra pars compacta loss with widespread α-synuclein–positive neocortical Lewy bodies in a patient with biallelic RFC1 AAGGG expansions with parkinsonism, dysautonomia, and cognitive decline. This indicates that Lewy pathology can co-occur in RFC1 disease despite often-negative ancillary synuclein biomarkers, underscoring mechanistic heterogeneity and the need for systematic clinicopathologic correlation. Taken together, these data warrant prospective, multicenter studies with standardized clinical scales and harmonized imaging protocols, systematic autonomic and sensory testing, and granular genotyping of RFC1 repeat configurations to delineate the true prevalence and pathophysiologic mechanisms of movement disorders across the RFC1 spectrum. Cognitive Impairment and Behavioral Changes in RFC1 Disease The pooled prevalence of cognitive symptoms was 31% among individuals with RFC1 disease. However, reported rates varied widely (0%–94%; 95% CI 13%–53%) and tended to be higher in studies primarily focused on cognitive/behavioral features of RFC1 disease,19,20,30,42 demanding caution in interpretation. The definition of cognitive impairment was inconsistent and, in most studies, undefined. Only 55 of 786 patients (7%) underwent a comprehensive battery of neuropsychological tests in 5 studies.18,20,30,32,42 Six studies8,11,17,28,37,39 used a single cognitive screening scale (either MMSE or MoCA), while the remaining 8 reported cognitive impairment (or the lack of) by clinical history/judgment, but no results of formal assessments were mentioned. MoCA is more sensitive to mild executive–-visuospatial deficits, whereas MMSE may show ceiling effects and miss subtle impairment.48 In RFC1 disease, where fronto-cerebellar network dysfunction is expected, MoCA may therefore outperform MMSE, which is better suited to hippocampal-predominant impairment. Thus, between-study differences may partly reflect the cognitive instrument used rather than true disease heterogeneity. All studies were case series (some of family members)18,30 and included no healthy control groups. Considering these methodological pitfalls and the lack of details on evaluation protocols and criteria to define deficits (e.g., the use of normative data matched for demographic variables, cutoff values, or published diagnostic criteria), cognitive impairment may have been overestimated in some studies, while being overlooked in others. As a consequence, the extent of the impact of RFC1 disease on cognition remains unclear, since the prevalence of cognitive impairment found in this meta-analysis (31%; 95% CI 13%–53%) differs from the global estimate (19%; 95% CI 5%–41%) in older adults of comparable age.49 Age was not associated with the risk of cognitive impairment, while it was more likely the longer the disease duration. This suggests that accumulation of neural damage, even beyond the cerebellum, may be required to cause clinically detectable cognitive decline, primarily in biallelic AAGGG expansion carriers. Many patients with cognitive symptoms had atrophy and/or hypometabolism in the cerebellum, but also in the brainstem17,35 and cortical areas,27,35 more often in the frontal lobes.18,19,38 While global cerebellar volume was positively correlated with phonemic fluency in 7 patients,20 to date, comprehensive investigations of neural correlates of cognitive symptoms in RFC1 disease and of the potential contribution of neurodegenerative comorbidities (e.g., PD, Alzheimer disease) are lacking. When different cognitive functions were assessed, primarily in investigations of CCAS, deficits most often involved executive function, attention, speed of information processing, and working memory, in line with clinical observations.19,35,38 This scenario is consistent with current knowledge of large portions of the posterior cerebellar lobes being connected with associative fronto-parietal areas that support control functions of cognition and behavior.50 Reports of long-term memory impairments are scarcer,30,32 possibly reflecting more limited cerebellar involvement in episodic memory.e1 Neuropsychiatric symptoms were less prevalent than cognitive deficits, suggesting that they may represent a clinically relevant issue only in a minority of patients.30 It is also possible that behavioral alterations due to progressive cerebellar neurodegeneration underlying RFC1 disease may be milder than those observed in CCAS caused by cerebrovascular lesionse2 and, therefore, may go unnoticed. Depression and anxiety were the symptoms most commonly reported, and the overall pattern of behavioral alterations was consistent with the domains included in the CCAS scale.e3 Prevalence estimates for specific neuropsychiatric features cannot be determined accurately with the published data available, since symptoms varied widely across the small subgroup of studies (10 of 37) with behavioral data, and details on assessments of symptom presence and severity were often missing. Similarly, the relative contribution of cerebral alterations highlighted, namely, cerebellar atrophy,27,38 brainstem atrophy,11,27 and putaminal dopaminergic dysfunction,37,38 to psychiatric phenotypes in RFC1 disease is yet to be clarified. Therefore, RFC1 disease may present with cognitive and/or behavioral symptoms in a subgroup of patients. Future investigations should clarify whether any social cognitive functions, which are also supported by cerebellar activity and relevant to behavioral control,50 may be vulnerable in this population. Study Limitations To ensure thoroughness, we did not exclude studies based on quality scores. However, data analysis proved challenging due to substantial heterogeneity, and the overall certainty of evidence (GRADE) was low. Methodological details were often incomplete, and when available, approaches varied widely, underscoring the need for standardized protocols. Most studies involved small cohorts, and several primarily investigated the prevalence of RFC1 expansions in PD, MSA, or MND populations. Many studies were phenotype-driven and did not systematically assess all clinical domains; therefore, estimates for individual features may be influenced by differential ascertainment across studies. Because our search strategy targeted atypical manifestations, their prevalence may have been overestimated, and publication bias remains a concern. Referral bias may result from including patients from movement disorder, MND, or cognitive clinics. Conversely, studies not designed to systematically assess atypical features may underestimate their frequency. Given the large multicenter nature of some cohorts, unreported overlap of participants across publications cannot be completely excluded; however, sensitivity analyses excluding the largest international cohorts yielded broadly overlapping CIs and did not materially change the overall interpretation. Finally, evidence for motor neuron, extrapyramidal, and cognitive involvement largely derives from cross-sectional data, limiting inferences on progression, and potential confounders (e.g., comorbidities) were rarely addressed. Conclusion Motor neuron signs, parkinsonism, and cognitive impairment may be part of the RFC1 disease spectrum, but their true prevalence remains uncertain due to heterogeneous findings. Subtle involvement may go unnoticed without systematic screening, yet these atypical features should not be overemphasized, and their contribution to patient disability remains to be fully understood. Larger multicenter studies with standardized protocols are needed to better delineate the phenotypic spectrum of RFC1 disease.