Clinical, Genetic, and Imaging Characteristics of SCA27B: Insights from a Large Dutch Cohort

Abstract

Background

Deep intronic GAA repeat expansions in intron 1 of the FGF14 gene were identified in 2023 as cause of late‐onset cerebellar ataxia. Since then, GAA‐FGF14‐related ataxia (SCA27B) has emerged as one of the most common genetic causes of late‐onset cerebellar ataxia.

Objectives

To describe the clinical, genetic, and imaging features of a large Dutch cohort.

Methods

The Radboudumc genetic database was queried for GAA‐FGF14 expansions ≥200. Repeat length was assessed using locus‐spanning polymerase chain reaction (PCR), repeat‐primed PCR, and PacBio sequencing. A subset was validated using Oxford Nanopore. Clinical and imaging data were retrospectively reviewed.

Results

127 individuals with GAA‐FGF14 expansions ≥200 were identified; clinical data were available from 116, including 109 symptomatic and 7 asymptomatic/presymptomatic individuals. Fifteen individuals carried GAA_200‐249 expansions; 60% exhibited at least one core SCA27B feature. Episodic symptoms occurred in 72.5%; 24% had prior emergency department or outpatient transient ischemic attack clinic visits. Brain magnetic resonance imaging frequently showed non‐specific white matter abnormalities (>90%); Superior cerebellar peduncle sign was present in 67.7%. Among those treated, 54.1% reported symptomatic benefit from 4‐aminopyridine. Nanopore and PacBio sequencing results showed high correlations. We observed an inverse relationship between age at onset and disease progression.

Conclusions

Core SCA27B features emerge in those carrying GAA_200‐249 expansions. SCA27B may mimic stroke in patients with episodic symptoms. In our cohort, later onset was associated with faster disease progression. The superior cerebellar peduncle sign may aid diagnosis, while relevance of white matter changes remains unclear. Positive response to 4‐aminopyridine was reported in approximately half of patients. © 2026 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

1. Introduction

In 2023, deep intronic GAA repeat expansions in intron 1 of the fibroblast growth factor 14 gene (FGF14) were identified as a frequent cause of late‐onset cerebellar ataxia (LOCA). ¹ , ² Since then, GAA‐FGF14‐related ataxia (SCA27B) has rapidly emerged as one of the most common genetic causes of LOCA, accounting for a substantial proportion of previously unexplained cases. Frequencies up to 65% have been reported in ataxia series, although with marked variability depending on ethnicity and inclusion of familial versus sporadic cases. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ In the Netherlands, GAA‐FGF14 expansions were previously identified in 11% of a cohort of patients with previously unsolved ataxia. ¹⁰

The pathogenic threshold of the GAA‐FGF14 repeat expansion remains uncertain. Uninterrupted GAA expansions of ≥300 repeats are considered highly penetrant, while those ranging from 250 to 299 repeats may show reduced penetrance. ³ , ⁴ , ¹¹ , ¹² More recently, alleles with 200–249 repeats have been proposed as potentially pathogenic, though supporting evidence remains limited. ³ , ¹³ , ¹⁴ , ¹⁵ , ¹⁶

SCA27B typically presents with (very) late‐onset, slowly progressive (axial) cerebellar ataxia, with or without episodic onset, and episodic symptoms during the disease course. ¹⁷ Abnormal eye movements such as downbeat nystagmus (DBN) and oscillopsia are common. ³ Symptomatic treatment with 4‐aminopyridine (4‐AP) has shown promising effects on ataxia severity and DBN in some studies, ³ , ⁸ , ¹⁷ but randomized trials are still lacking.

Distinctive imaging findings include T2‐hyperintensities along the superior cerebellar peduncle (SCP) and its decussation. ¹⁸ Less specific features such as cerebellar atrophy and nonspecific cerebral white matter changes have also been reported. ⁸ , ¹⁷ , ¹⁸ , ¹⁹

We present the phenotypic profile, imaging findings, and response to 4‐AP in a large cohort of 109 Dutch ataxia patients with GAA‐FGF14 repeat expansions ≥200 GAA repeats.

2. Methods

We queried the Radboudumc genetic testing database for diagnostically reported intronic GAA‐FGF14 repeat expansions in ataxia patients or family members seen in neurology/genetics departments of Radboudumc or affiliated Dutch centers since September 2023. Although the currently accepted diagnostic threshold for pathogenicity is 250 GAA repeats, we extended the threshold for this study to 200 GAA repeats to investigate the hypothesis that a threshold closer to 200 repeats might also be clinically relevant. ¹⁴ For external patients, treating neurologists/geneticists provided clinical data.

2.1. Phenotyping

Clinical records were systematically assessed according to a prespecified data sheet, including demographic data (sex, age at onset [episodic or permanent ataxia], age at assessment, disease duration), family history, comorbidity, neurological phenotype (episodic symptoms, emergency department [ED] visit related to episodic symptoms, presence of oculomotor signs, upper motor neuron findings, movement disorders, neuropathy, autonomic/cognitive/neuropsychiatric features), 4‐AP use, and disease severity. Disease severity was expressed by the Scale for the Assessment and Rating of Ataxia (SARA) score ²⁰ when available, as well as the Friedreich Ataxia Rating Scale Functional Disability Stage (FARS‐FDS). ²¹ Longitudinal clinical data on SARA score were available for 10 individuals with a follow‐up to 6 years.

2.2. Treatment with 4‐AP

The effect of 4‐AP was retrospectively assessed (use, ongoing treatment, perceived effect, domains impacted, side effects). Effects were categorized as positive, neutral, or negative.

2.3. Genetic Analysis

GAA‐FGF14 repeat expansion length was analyzed using fragment length analysis of locus spanning polymerase chain reaction (PCR), repeat‐primed PCR, and PacBio sequencing of long‐range PCR fragments. Detailed descriptions of the genetic and sequencing methodologies are provided in the Supporting Information (S1). A subset of samples (n = 92) was sent to the Institute of Neurogenetics, University of Lübeck for additional Nanopore sequencing to test the robustness and compare sensitivity across two long‐read technologies. This technique has been previously described in detail. ¹² The scripts and reference file are provided at: https://github.com/joshua21997/FGF14-repeat-expansion.

Additional testing excluded alternative causes in individuals with a GAA_200‐249 expansion (n = 15), including expansions in ATXN1‐2‐3, CACNA1A, ATXN7, TBP, DRPLA (n = 13), RFC1 (n = 13), FMR1 (n = 1), FXN (n = 7), point mutations in SPG7 (n = 7), and a whole exome sequencing gene panel for movement disorders in eight patients.

Genetic testing was performed on clinical diagnostic request with informed consent. The study complied with Radboudumc regulations and was approved by the ethics board (niet‐WMO verklaring, 2024‐17849).

2.4. MRI Analysis

For 64 individuals magnetic resonance imaging (MRI) scans were available for review, including sagittal three‐dimensional (3D) T1‐ and 3D FLAIR (fluid‐attenuated inversion recovery) T2‐weighted images. Scans were reviewed by one of the authors using a standardized protocol adapted from Chen et al. ¹⁹ focusing on (1) the presence/absence and degree of atrophy of the vermis and cerebellar hemispheres, (2) the presence/absence of the SCP sign and its intensity (faint/prominent), (3) the presence/absence of supratentorial atrophy and its degree, and (4) the presence/absence of white matter abnormalities (WMAs) and Fazekas score for periventricular and deep white matter (0–3) as well as Fazekas sum score (0–6). Age was considered when rating atrophy severity. The reviewer was blinded for genetic/clinical status. Supplementary longitudinal scans were available in 21 individuals. The most recent scan was used, while severity/progression of white matter changes was compared with the earliest available scans.

2.5. Statistical Analysis

Between‐group differences (GAA_200‐249, GAA_250‐300, or GAA_≥300) in continuous variables (SARA/FARS, onset age) were tested with Kruskal–Wallis, and categorical variables (DBN, episodic symptoms) with a chi‐squared test.

2.6. Clinical Associations

The relationship between GAA‐FGF14 repeat length and age at onset, disease severity (SARA score or FARS stage), and cross‐sectional disease progression (SARA score or FARS stage divided by self‐reported disease duration) was assessed using linear regression analysis. Age at onset, SARA/FARS scores, or disease progression estimates were treated as dependent variables and GAA‐FGF14 repeat length was included as the predictor. The analyses were adjusted for relevant covariates: disease duration and age at assessment for models involving absolute SARA/FARS scores, and age at assessment and age at onset for models assessing disease progression.

In 10 cases with longitudinal data available, SARA progression was modeled with linear mixed‐effects (lme4 in R), including age and duration as fixed effects and subject as random intercept.

For radiological associations, between‐group differences for categorized groups based on severity of atrophy or presence of SCP sign were compared for GAA‐FGF14 repeat length and age at onset using a nonparametric Kruskal–Wallis test. Associations between severity of WMAs, GAA‐FGF14 repeat length, and age (at scanning) were analyzed with linear regression with both periventricular Fazekas scores and Fazekas sum score as the dependent variable and GAA length and age (at scanning) as predictors.

All analyses were conducted in Rstudio (2024.04.2).

3. Results

3.1. Characteristics of Participants

Some 127 subjects with GAA‐FGF14 repeat expansions exceeding the threshold of 200 repeats were identified through diagnostic testing. Clinical data were available for 116 subjects. Seven were asymptomatic/presymptomatic, leaving 109 symptomatic for phenotype analysis (S2). Thirty‐seven had used 4‐AP. Median age was 71 (range 40–86) years, disease duration 8 (range 0.5–42) years, and 40% were female. The median onset age was 61 (range 25–80) years.

3.2. Genetic Characteristics

Fifteen patients had GAA‐FGF14 _200–249 expansions, 18 had GAA‐FGF14 _250–299, and 76 had GAA‐FGF14 _≥300 (Fig. 1). In six individuals with GAA_200–249 expansions pathogenicity remained uncertain because none of them had a phenotype suggestive of SCA27B. These cases were excluded from genotype–phenotype correlation analyses. Three of them had repeat lengths at the lower end of our defined ≥200 repeat threshold (see S3 for a GAA length comparison between uncertain cases and those classified as likely pathogenic).

FIG. 1.

Distribution of symptomatic FGF14‐GAA_>200 expansion carriers within genetic subgroups based on GAA length and frequency of core SCA27B feature determined by presence of either downbeat nystagmus (DBN), episodic symptoms, or superior cerebellar peduncle (SCP) involvement on magnetic resonance imaging (if available). [Color figure can be viewed at wileyonlinelibrary.com]

A (probable) positive family history was reported in 59.3% of patients. Among these, maternal transmission was confirmed or strongly suspected in 32.1%, and paternal transmission in 15.6%. In the remaining patients, the route of transmission was unclear. Segregation data were available for seven individuals, revealing paternal contraction in one individual and expansion in six (five of whom showed maternal transmission), including transmission of an intermediate ±200 allele by a patient with a biallelic (>300/200) expansion further expanding in her offspring. In individuals with GAA_200‐249 repeats, no segregation data were available, but 40% of these individuals had a positive (or suggestive) family history.

A strong correlation was observed between PacBio and Nanopore sequencing methods (R: 0.93, P < 0.0001) (Supporting Information S4), particularly for GAA repeat lengths between 200 and 400. However, for repeat lengths exceeding 400 GAA units (n = 11), Nanopore sequencing tended to report longer repeat lengths than PacBio sequencing.

3.3. Phenotypic Profile

Nearly all symptomatic individuals had chronic gait ataxia; 54% also showed dysarthria. One patient presented with episodic ataxia. Median SARA score was 7.0 points (range 0–23), median FARS disability stage was 2 (range 1–4.5) and median estimated cross‐sectional rate of SARA progression was 0.9 points/year.

Frequencies of other clinical features are shown in Figure 2A. Abnormal eye movements were most common, especially downbeat (DBN) and/or gaze‐evoked (GEN) nystagmus (any type: 69%, DBN: 38%, GEN: 35%). Dystonia (14.7%) and dystonic tremor (13.7%) were the most frequent additional movement disorders. Upper motor neuron findings were present in 13.1%. Neuropathy (clinical or electromyography confirmed) occurred in 17%.

FIG. 2.

(A) Frequency of clinical features in FGF14‐GAA_>200 expansion carriers. (B) Donut plot showing the frequency and type of episodic or fluctuating symptoms. [Color figure can be viewed at wileyonlinelibrary.com]

Seventy‐five patients (72.5%) experienced episodic or fluctuating symptoms (Fig. 2B), and in at least 31.2% these preceded the onset of ataxia. Vertigo (n = 39) was most frequently reported, followed by (gait) ataxia (n = 30), dysarthria (n = 26), and diplopia (n = 18). Several subjects exhibited multiple different features. Among those with episodic symptoms, 18 (24%) had visited the ED or transient ischemic attack (TIA) outpatient clinic during their disease course, but prior to the diagnosis of SCA27B, and were initially suspected to have suffered from vertebrobasilar infarction or TIA or peripheral vestibular syndrome. Other differential diagnoses (not limited to ED visit) included functional neurological disorder and myasthenia gravis.

We defined the presence of “core SCA27B features” by the occurrence of at least one of the following: episodic symptoms, DBN, or SCP sign on brain MRI (if available). Among subjects with GAA_200‐249, 60% presented with at least one core feature, and six individuals (40%) exhibited all three features. In the group with GAA_250‐299, 83.3% exhibited at least one core feature, with an even higher proportion observed in subjects with GAA_≥300 repeats (88.2%; Fig. 1).

Some 22.4% of the symptomatic individuals required a walking aid or were wheelchair‐dependent, with a median disease duration of 9.5 years in the subgroup of patients requiring a walking aid (FARS disability stage = 3). In 14 patients, cross‐sectional SARA progression was above the interquartile range (IQR) of 2 points/year. In these individuals, autonomic features were present in 57% (8/14) and included mainly urinary symptoms and orthostatic hypotension.

3.4. Genotype/Phenotype Correlations

Age at onset inversely correlated with GAA‐FGF14 length (P = 0.002, R² = 0.09), indicating earlier onset with longer repeats (Fig. 3A).

FIG. 3.

(A) Association between age at onset and GAA length. (B) Association between cross‐sectional Scale for the Assessment and Rating of Ataxia (SARA) progression, age at onset, and GAA length. (C) Association between cross‐sectional Friedreich Ataxia Rating Scale (FARS) progression, age at onset, and GAA length. The boxes indicate the (scaled) estimates and P‐values of a linear regression model including SARA/FARS progression as dependent variable and age at assessment, age at onset, and GAA length as predictors, as well as overall model performance. [Color figure can be viewed at wileyonlinelibrary.com]

We assessed SARA scores and progression against GAA‐FGF14 length, onset age, disease duration, age at assessment (S6A), and sex (S6B). Progression correlated with duration and onset age, and negatively with GAA‐FGF14 length. No significant associations with age at assessment and no differences between female or male patients in progression rate were observed.

We next modeled the relationship between SARA scores and GAA repeat length, including disease duration as a covariate. A statistically significant inverse association between GAA repeat length and SARA (scaled estimate: –1.8, P = 0.007), alongside a positive association with disease duration (scaled estimate: 2.1, P = 0.001), suggested that patients with shorter GAA alleles or longer disease duration tend to exhibit more severe ataxia (adjusted R² = 0.23). When age at assessment was added as an additional covariate, it emerged as a significant predictor of SARA (scaled estimate: 1.7, P = 0.02), and the effect of GAA repeat length was attenuated (scaled estimate: –1.09, P = 0.11). The overall explanatory power of the model improved (adjusted R² = 0.32), suggesting that age‐related factors may contribute independently to disease severity and partially confound the relationship between GAA length and ataxia severity.

Moreover, we examined factors associated with disease progression. A significant inverse association between GAA repeat length and progression rate (scaled estimate: −0.31, P = 0.03) was detected in a model including GAA length and age at assessment, suggesting that individuals with shorter alleles tend to progress more rapidly. Age at assessment was not a significant predictor (P = 0.83) and the overall explained variance was modest (adjusted R² = 0.1). When adding age at onset to the model, it appeared as a strong significant predictor of progression (scaled estimate: 0.64, P = <0.001) with overall improved fit of the model (adjusted R² = 0.34), while GAA‐FGF14 length was no longer a significant independent predictor in this model (P = 0.4) (Fig. 3B).

When analyzing FARS progression, a similar trend was observed: shorter GAA repeats were associated with faster progression (scaled estimate: −0.085, P = 0.06), while age at assessment was not significantly associated. After addition of onset age to the model, overall fit improved (adjusted R² = 0.37) and age at onset was a strong significant predictor of disease progression (scaled estimate: 0.22, P = <0.0001) (Fig. 3C).

Longitudinal SARA scores were available in 10 patients. Mixed‐effect modeling in this subset estimated SARA progression at 0.3 points/year (95% CI: 0.03–0.55). No significant interaction between GAA length and disease duration (suggesting effect on progression) was detected in this small subset.

Median SARA and FARS appeared higher in the 200–249 repeat subgroup, but no significant differences in scores or progression were found across groups (GAA_200‐249, GAA_250‐299, GAA_≥300). Onset age and phenotypic features (DBN, episodic symptoms) also did not differ.

3.5. Brain MRI

Of 64 individuals with available routine brain MRI (median age at scanning: 67 [IQR: 63–74] years), 53.2% had mild vermis atrophy and 30.6% had moderate vermis atrophy. Furthermore 53.8% had mild atrophy and 6.2% had moderate atrophy of the cerebellar hemispheres. Some 67.7% of individuals had faint or prominent hyperintensity of the SCP or decussation of the SCP in the midbrain, in the majority of cases more pronounced on T2‐weighted imaging rather than FLAIR‐T2 (only 35.5%). No statistically significant differences in GAA repeat length or age at onset were observed between groups categorized by SCP sign (absent, faint, or prominent) or by vermis/cerebellar hemisphere atrophy (none, mild, moderate). No SCP sign was present in individuals with a GAA_200–249 expansion and uncertain pathogenicity (4 of 6 MRIs available for review). Supporting Information S5 shows examples of a prominent and faint SCP sign in two individuals. Supratentorial atrophy was observed in 40.6% and was mainly mild (34.4%).

WMAs were common (94%), mainly with a periventricular location. Fazekas score >2 occurred in 52% (periventricular) and 28% (deep white matter); median sum score was 3 (IQR 1–4). Some 51.6% of patients had a periventricular Fazekas score > 2 and 28.1% had a deep white matter Fazekas score > 2. The median sum Fazekas score was 3.0 (IQR:1–4). Four individuals aged <50 years (including one pre‐symptomatic person) showed WMAs. Vascular risk factors were identified in 61.7% of the subjects with WMAs. Figure 4B shows example WMAs in relatively young subjects with no vascular risk factors. Twenty‐one subjects had longitudinal MRI scans available, and WMAs were already present on the initial brain MRI. WMAs were visually unchanged or minimally progressive in 19 subjects, but Fazekas scores were comparable between scans, with up to 6 years between the earliest and last scans. Linear regression analysis between both periventricular Fazekas scores as well as Fazekas sum score and GAA length, including also age as a predictor, revealed statistically significant associations with both age and GAA length (Fig. 4A).

FIG. 4.

(A) Association between Fazekas sum score and age, coloring indicates GAA length in all subjects. The box indicates the (scaled) estimates and P‐values of a linear regression model including Fazekas sum score as dependent variable and age and GAA length as predictors, as well as overall model performance. (B) Magnetic resonance imaging examples showing white matter abnormalities (WMAs) in relatively young subjects with no vascular risk factors. [Color figure can be viewed at wileyonlinelibrary.com]

3.6. 4‐AP Response

Thirty‐six patients had used or were using 4‐AP at the time of inclusion; 52.8% of subjects reported a positive effect, while 28% reported a neutral or negative effect. In all other individuals, the effect was unclear or not yet evaluated. Symptomatic improvement was reported for DBN or oscillopsia (n = 8), gait (n = 12), and dysarthria (n = 8). Six patients quit 4‐AP use due to side effects.

4. Discussion

This study represents one of the largest reported cohorts of SCA27B worldwide. Our cohort confirms established hallmarks of SCA27B, supports more recent findings, and provides relevant new insights. First, we provide additional evidence supporting the pathogenicity of GAA alleles in the 200–249 repeat range. Second, individuals with episodic presentation frequently visit the ED before diagnosis. Third, while confirming prior imaging hallmarks, our results further support WMAs as a common neuroradiological feature of SCA27B. Fourth, we demonstrate that Nanopore and PacBio sequencing methods yield comparable results. Finally, an inverse relationship between age at onset and disease progression was identified, suggesting a faster disease progression in those with older onset age.

Although the exact prevalence needs to be elucidated, our data confirm that GAA‐FGF14 repeat expansions are a common cause of LOCA in the Netherlands, ¹⁰ with 127 GAA_>200 expansions found within 2 years of test availability. Fifteen patients carried GAA_200‐249 expansions. Five studies have suggested pathogenicity of GAA_200‐249 expansions by showing enrichment of these alleles in DBN³ and ataxia ¹³ , ¹⁴ , ¹⁵ , ¹⁶ patients relative to controls and four studies showed segregation with disease in families with LOCA. ⁹ , ¹¹ , ¹³ , ¹⁴ although with incomplete penetrance. We identified nine individuals within this group with at least one of the core SCA27B features, while six had all core SCA27B features. In six other patients, uncertainty about the pathogenicity of the GAA‐FGF14 repeat expansion remained because of an incompatible phenotype. Three of them had GAA repeat lengths close to the predefined threshold of 200 repeats, in line with the hypothesis that the pathogenic threshold may be slightly higher than 200 uninterrupted GAA repeats. ¹⁴ In the other cases, other factors not considered in this study may have influenced pathogenicity, such as mosaic interruptions within the GAA repeat sequence. ¹²

Interestingly, 24% of patients with episodic features presented to the ED prior to their SCA27B diagnosis and were initially suspected of having a vertebrobasilar infarction or TIA, confirming that SCA27B can present as a stroke mimic. Therefore, in elderly patients presenting with (recurrent) acute vertigo and negative diffusion‐weighted MRI, accompanied by slowly progressive (mild) gait disability or DBN, SCA27B should be considered in the differential diagnosis.

The previously reported SCP sign ¹⁸ was present in almost 70% of subjects, particularly on T2‐weighted images, confirming its potential as a diagnostic clue. As previously noted, ¹⁹ non‐specific WMAs were also common (93.8%), with periventricular Fazekas scores >2 observed in 51.6% of subjects. This suggests an increased prevalence compared with WMAs rates in healthy elderly individuals of comparable age in the Netherlands. ²² WMAs correlated with both age and GAA repeat length, indicating that these abnormalities might be a disease‐specific feature rather than solely attributable to aging or small vessel disease. Besides, WMAs were also stable and present on longitudinal imaging, appearing on scans up to 6 years prior to diagnosis, which supports the hypothesis that WMAs might be an early radiological characteristic of SCA27B. Nevertheless, in almost 62% of the individuals with WMAs, vascular risk factors such as hypertension and smoking were identified, and might point towards small vessel disease as the primary etiology. Vermis and cerebellar atrophy were frequent but less specific for diagnosis.

Unexpectedly, we found a significant association between disease progression and age at onset, suggesting faster progression in patients with older age at onset. Furthermore, an inverse association between GAA‐FGF14 repeat length and progression was also identified, suggesting that patients with shorter GAA‐FGF14 expansions may experience faster disease progression. However, GAA repeat length was no longer independently associated with progression in a model alongside age at onset, suggesting that the effect of age at onset may only be partially mediated by GAA repeat length. Several explanatory factors need to be considered. First, methodological factors may have partly influenced our results. Disease progression was estimated cross‐sectionally using the current SARA score in relation to self‐reported disease duration. The mean progression rate calculated by this method was relatively high, at 0.9 points on SARA per year, indicating that this approach may overestimate the true rate of disease progression. Limited longitudinal data on disease progression were available, but our estimates of disease progression were similar to the previously reported rate of 0.3 to 0.4 points/year. ⁷ , ¹⁷

Second, the inverse correlation between age at onset and GAA‐FGF14 repeat length found in our cohort suggests that individuals with shorter GAA expansions tend to develop symptoms later in life. Similar patterns have been observed in other neurodegenerative diseases, such as Parkinson's disease (PD), where late onset has been associated with faster progression ²³ , ²⁴ mainly within the gait domain, compared with slower motor progression observed in early‐onset PD. Whether the apparent acceleration of progression in individuals with older age at onset is biologically mediated by genetic factors such as GAA‐FGF14 repeat length, or instead reflects age‐related factors or comorbidities typical of very late‐onset disease, remains to be determined.

Finally, it is possible that patients with shorter GAA repeats (GAA_<250) might have an alternative (genetic) diagnosis accounting for the more rapid progression; however, alternative known genetic causes were thoroughly excluded in most of them, and other genetic ataxias do typically present with younger onset age. GAA‐FGF14 repeat length has previously been suggested as a modifier of disease course in the cerebellar type of multiple system atrophy (MSA‐C), ²⁵ with faster progression observed in individuals carrying (intermediate) GAA‐FGF14 expansions. Co‐pathology might also have been of influence in our cohort, as several patients with faster progression did show autonomic dysfunction, which might be compatible with an MSA‐C phenotype, though severity of autonomic symptoms was not systematically assessed.

GAA repeat expansion length was determined using both Nanopore and PacBio sequencing, showing a strong correlation between the two methods. Mainly in FGF14 alleles with more than 400 GAA repeats, Nanopore tended to report longer repeat lengths than PacBio. If GAA repeat length is confirmed as a modifier of age at onset or disease progression, these differences may become diagnostically or clinically relevant.

The positive treatment effect of 4‐AP observed in approximately half of the patients in our cohort was comparable to that reported in previous retrospective studies. ³ , ¹³ , ¹⁷ Although a significant proportion of patients already received 4‐AP at the time of inclusion, there remains a high unmet need for randomized controlled trials or other prospective clinical studies with objective endpoints to definitively establish its efficacy in SCA27B.

This study has several limitations. The cohort reflects the diagnostic yield of SCA27B‐positive individuals identified during routine clinical care. Consequently, deep phenotyping and standardized assessments of disease severity or MRI measures could not be performed, as we were limited to information available in clinical records and to neuroimaging obtained for diagnostic rather than research purposes. Similarly, data on 4‐AP efficacy were retrospectively extracted from clinical records and could not be assessed systematically.

In conclusion, we describe a large diagnostic cohort of SCA27B patients, confirming the significance of GAA‐FGF14 repeat expansions as a cause of LOCA in the Netherlands. Although the phenotype is highly recognizable to clinicians who are aware of it, SCA27B can present as a stroke mimic in some cases. MRI features, especially the SCP sign, may aid in diagnosis. The role and specificity of WMAs in SCA27B warrant further study, as does the relationship between age at onset, GAA‐FGF14 repeat length, and disease progression.

Author Roles

(1) Research Project: A. Conception, B. Design, C. Data Acquisition; (2) Statistical Analysis: A. Design, B. Data Analysis, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Editing the Final Version.

B.v.d.W., T.H.v.P., E.‐J.K.: 1A, 1B.

T.H.v.P., M.P., R.P.P.W.M.M., J.d.V., C.V.‐B., V.O., S.K.L.D., M.H., M.O., A.B., J.v.d.W., E.V., T.H.W., L.K., E.d.B., J.v.G., M.B., D.S.R., J.I.H., T.C., M.S., T.G., E.v.d.L., C.K., J.T., B.v.d.W.: 1C.

T.H.v.P., M.P.: 2B.

T.H.v.P.: 3A.

M.P., R.P.P.W.M.M., J.d.V., C.V.‐B., V.O., S.K.L.D., M.H., M.O., A.B., J.v.d.W., E.V., T.H.W., L.K., E.d.B., J.v.G., M.B., D.S.R., J.I.H., T.C., M.S., T.G., E.v.d.L., C.K., J.T., E.J.‐K., B.v.d.W: 3B.

Supporting information

Supporting Information S1 Genetic and sequencing methodology.

Fragment length analysis of locus‐spanning polymerase chain reaction (PCR) was determined using synthetic fluorescently‐labeled primers (Forward: 5′‐FAM‐ AGCAATCGTCAGTCAGTGTAAGC‐3′ and Reverse 5′‐CAGTTCCTGCCCACATAGAGC‐3′) flanking the GAA repeat region. The triplet repeat primed PCR (TP‐PCR) was performed using two gene specific primers that lie outside the repeat (P1: 5′FAM‐TCCCTAGTGTCTGCTGGCTAA‐3′ and P3: 5′‐CAGGAAACAGCTATGACC‐3′) and a third primer designed across the repeated sequence (P4: 5′‐CAGGAAACAGCTATGACCCTTCTTCTTCTTCTTCTTCTT‐3′). PCR amplification conditions included initial denaturation of 95°C for 10 min then 35 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s with extension at 72°C for 1 min and a final extension at 72°C for 7 min. This was followed by direct analysis of the length of the amplified products by capillary electrophoresis using the LifeTechnologies 3730 XL Analyzer (Thermo Fisher). Analysis of results was performed using Genemarker V2.6.7 (SoftGenetics LLC). When fragment length analysis and repeat‐primed PCR showed a (possibly pathogenic) expansion, long‐read sequencing of long‐range PCR fragments was performed to confirm the repeat length and validate the pathogenic sequence (ie, GAA expansion rather than GAAGGA). Long‐range PCR was performed using two primers (Forward: AGCAATCGTCAGTCAGTGTAAGC and Reverse CAGTTCCTGCCCACATAGAGC) and LongAmp HotStart Taq mastermix (M0533S, Bioke). PCR conditions comprised an initial denaturation at 94°C for 30 s, then 25 cycles of denaturation at 94°C for 30s and extension at 60°C for 5 min, and a final extension at 65°C for 10 min. Long‐read sequencing of the PCR fragments was done on a PacBio Sequel IIe system (Pacific Biosciences) as described (PMID: 36351915). Briefly, LR amplicons were purified by AMPure PB beads (Pacific Biosciences), using a bead ratio of 1.5×. Library preparation was done according to protocol ‘Procedure and Checklist—Preparing SMRTbell Libraries using PacBio Barcoded Adapters for Multiplex SMRT Sequencing’ (Pacific Biosciences, Part Number 100‐538‐700‐02). Generation of polymerase‐bound SMRTbell complexes was performed using the Sample Setup option in SMRTLink (Pacific Biosciences). Sequencing was performed using the Run Design option in SMRTLink. Libraries were loaded using diffusion loading with an on‐plate concentration of 4.5 pM. All runs were sequenced using a movie time of 20 hr per SMRTcell and included pre‐extension. Coverage depth typically exceeds 10,000 reads, with a bias to the shorter (normal) allele.

S2. Study flow.

S3. Distribution of GAA length in individuals within a GAA_200‐249 repeat expansion.

S4. Correlation between GAA length determined by PacBio versus Nanopore sequencing.

S5. Examples of a (A) prominent (T2‐FLAIR) and (B) faint superior cerebellar peduncle sign (T2) in individuals.

S6A. Relationship between cross‐sectional disease progression and (A) GAA length, (B) age at onset, (C) age at assessment, and (D) disease duration.

S6B. Cross‐sectional disease progression by sex.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.