Neural Correlates of Intellectual Dysfunction in Episodic Ataxia Type 2

Abstract

Background and Purpose

Episodic ataxia type 2 (EA2) is characterized by episodes of vertigo and ataxia due to mutations in CACNA1A that encodes the α1A subunit of the P/Q-type voltage-gated calcium channel. This study aimed to identify neural correlates of cognitive dysfunction in EA2 by investigating brain atrophy in these patients and determining the relationships between regional brain volumes and intellectual dysfunction.

Methods

We recruited 12 patients with EA2 (including 6 males; age=30.5±14.3 years, mean±standard deviation) in 3 university hospitals of South Korea from 2019 to 2023. Regional brain volumes were quantified using voxel-based morphometry and the brainstem-structures feature of FreeSurfer. The results were compared with those for healthy controls. The relationships between regional gray-matter volumes (GMVs) and cognitive function were assessed using voxel-wise multiple regression analyses within the general linear model framework.

Results

Brain volumetry revealed a significant decrease in cerebellar volumes, particularly in the vermis (lobules IV, V, and VIII), bilateral flocculi (lobule X), and brainstem. The Full-Scale Intelligence Quotient was positively correlated with the GMVs in the left parahippocampal gyrus, right caudate nucleus, and right cerebellar crus II; the Verbal Comprehension Index was correlated with the GMVs in the bilateral cerebellar crura I and II; and the Processing Speed Index was correlated with the GMV in the right parahippocampal gyrus.

Conclusions

Volumetric analyses revealed brain atrophy in patients with EA2 that was correlated with the clinical features observed in this disorder. These findings may further expand the imaging spectrum of disorders associated with CACNA1A mutations, although the identified correlations need to be interpreted with caution.

Graphical Abstract

INTRODUCTION

Episodic ataxia type 2 (EA2) is characterized by episodic vertigo and ataxia due to mutations involving CACNA1A.¹,²,³ CACNA1A is located on chromosome 19p and encodes Cav2.1, the pore-forming α1A subunit of the P/Q-type voltage-gated calcium channel.¹,⁴ These channels are predominantly expressed on Purkinje and granule cells in the cerebellum.¹ Indeed, patients with EA2 show interictal ocular motor findings indicative of cerebellar dysfunction, which include downbeat nystagmus, gaze-evoked nystagmus, positional nystagmus, impaired smooth pursuit, and saccadic dysmetria.² The phenotypes of disorders due to mutations involving CACNA1A have recently expanded to include epilepsy and intellectual disability in addition to intermittent or progressive cerebellar dysfunction.⁵,⁶ These findings are consistent with these ion channels being expressed ubiquitously in both the central and peripheral nervous systems.¹ However, there are no previous reports on volumetric analyses of brain MRI data and their correlations with clinical features in EA2. This study aimed to identify alterations in brain volume and their correlations with intellectual dysfunction in patients with EA2.

METHODS

Patients

We recruited 12 patients with EA2 (including 6 males) in 3 university hospitals of South Korea from 2019 to 2023. The age of the patients ranged from 10 to 52 years, with a mean of 30.5 years (standard deviation [SD]=14.3 years, median=28.5 years). All patients showed clinical features of episodic vertigo and imbalance lasting hours to days, interictal eye movement abnormalities indicative of cerebellar dysfunction, and responses to acetazolamide. All patients also carried a mutation involving CACNA1A and had a genetically confirmed diagnosis of EA2. The clinical and genetic features of the patients had been reported previously, and the core features are summarized in Supplementary Table 1 (in the online-only Data Supplement).⁶

Comparisons were performed by including 19 healthy controls with comparable group-level age (t-test, p=0.795) and sex (chi-square test, p=0.667) distributions. The control group had an age of 30.7±5.8 years (mean±SD) and consisted of 8 males and 11 females.

Assessments of intellectual function

Intellectual function was assessed using the Korean version of the Wechsler Adult Intelligence Scale—Fourth Edition (WAIS-IV) or the Korean version of the Wechsler Intelligence Scale for Children—Fourth Edition (WISC-IV).⁷,⁸ The WAIS-IV and WISC-IV consist of the following subtests: Verbal Comprehension Index (VCI; core subtests: similarities, vocabulary, and information; supplemental subtest: comprehension), Perceptual Reasoning Index (PRI; core subtests: block design, matrix reasoning, and visual puzzles; supplemental subtests: picture completion and figure weights), Working Memory Index (WMI; core subtests: digit span and arithmetic; supplemental subtest: letter-number sequencing), and Processing Speed Index (PSI; core subtests: symbol search and coding; supplemental subtest: cancellation).⁷ The Full-Scale Intelligence Quotient (FSIQ) is the overall intelligence level of an individual, and was calculated based on the 10 core subtests.⁷ The FSIQ in the WAIS-IV typically has a mean of 100 and an SD of 15,⁷ meaning that 68% of people have an intelligence quotient within 1 SD of the mean (i.e., 85 to 115). The WAIS-IV classifies intelligence quotients into different levels: very superior (above 129), superior (120–129), high average (110–119), average (90–109), low average (80–89), borderline (70–79), and intellectual dysfunction (below 70).⁷ The Korean versions of the WAIS-IV and WISC-IV were administered in paper-and-pencil or digital form, with it typically taking 70–100 minutes to complete the core subtests.⁸

Acquisition of MRI data

In addition to routine T1-weighted, T2-weighted, and FLAIR (fluid-attenuated inversion-recovery) axial MRI as well as T1-weighted coronal MRI, T1-weighted high-resolution sagittal MRI of the whole brain was performed using a 3.0-T unit (Ingenia Elition, Philips) with the following parameters: echo time, 4.60 ms; repetition time, 8.30 ms; flip angle, 8°; number of excitations, 1; resolution, 1×1×1 mm3; and matrix, 240×240×175.

Voxel-based morphometry

Voxel-based morphometry (VBM) was conducted according to a previously described methodology⁹,¹⁰ with a minor modification to identify gray matter (GM) regions exhibiting atrophy in EA2 patients using an unbiased, whole-brain approach, and to identify the specific GM regions that are correlated with measures of intelligence. The analyses were performed using the CAT12 toolbox (version 12.5, build 1355; an extension of SPM12; https://neuro-jena.github.io/cat/) running on MATLAB R2019a. T1-weighted images were preprocessed using default CAT12 settings, including bias correction, noise reduction, and segmentation into GM, white matter, and cerebrospinal fluid. Segmented images were spatially normalized to a DARTEL (Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra) template in Montreal Neurological Institute (MNI) space that was modified to preserve the total GM volume (GMV) and smoothed with an 8-mm FWHM (full width at half maximum) Gaussian kernel.

All voxel-wise statistical analyses were performed within the general linear model (GLM) framework as implemented in SPM12. Within this framework, both the between-group comparisons (two-sample t-test) and the correlation analyses (multiple regression) were specified as distinct instances of the same GLM, with the modified GMV at each voxel serving as the dependent variable. For each model, parameter estimates (β values) for the regressors of interest and nuisance covariates were computed using ordinary least squares, and voxel-level t-statistics were derived from contrasts of these parameter estimates. For group comparisons, voxel-wise GM differences between EA2 patients and healthy controls with comparable age and sex distributions were assessed using two-sample t-tests in SPM12, with age and the total intracranial volume (ICV) as covariates. The criterion for statistical significance was set at p<0.05 (familywise error [FWE]-corrected), with a cluster-extent criterion of >50 contiguous voxels.

Analyses of the correlations between the regional GMVs and intelligence measures in EA2 patients were performed using a multiple regression GLM in SPM12. The FSIQ, VCI, PSI, PRI, and WMI were used as the covariates of interest, with age and the total ICV as nuisance covariates. Each model included one cognitive index as the independent variable of interest, with the modified GMV at each voxel as the dependent variable (residual degrees of freedom=8). Correlations were initially assessed at p<0.05 (FWE-corrected), but since no significant clusters survived this stringent criterion, positive correlations were assessed using an uncorrected criterion of p<0.001 with a cluster-extent criterion of >50 voxels. This combination of a voxel-level criterion and minimum cluster size is widely adopted in VBM studies of rare neurological conditions with small samples.¹¹,¹²,¹³ Significant clusters were anatomically labeled using the Neuromorphometrics atlas implemented in CAT12, with the peak coordinates reported in MNI space.

To substantiate the reported associations beyond the voxel-level criterion, effect sizes were quantified as Pearson coefficients (r values) for the correlations between the cognitive indices and the mean GMV extracted from each significant cluster. The GMV was extracted from smoothed GM probability maps using the cluster masks defined by the SPM12 regression analysis. Standardized residuals were computed after regressing out age and total ICV from both the cognitive indices and the regional GMVs prior to performing the correlation analyses, following the partial correlation approach as described previously.

Measurements of regional brain volumes

VBM primarily targets the cerebral and cerebellar GM and may not adequately capture the brainstem structures. We therefore utilized the Brainstem Substructures tool of FreeSurfer (version 6.0.0)¹⁴ to segment the brainstem into the midbrain, pons, medulla oblongata, and superior cerebellar peduncle, thus allowing more-detailed volumetric assessments in the EA2 patients. The brainstem volume and its ratio to the ICV were compared with the values of the healthy controls with comparable age and sex distributions using independent-samples t-tests with Welch’s correction.

Ethical considerations

All experiments followed the tenets of the Declaration of Helsinki. The study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. B-2407-911-101), who waived the requirement to obtain informed consent due to the retrospective observational design of the study. This observational study was conducted and reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. The observational design of this study meant that there is no clinical trial registration to report.

RESULTS

Intellectual dysfunction

Ten (83.3%) of the 12 patients showed below-average FSIQs (Table 1): low average (80–89) in 5 (41.7%), borderline (70–79) in 1 (8.3%), and intellectual dysfunction (below 70) in 4 (33.3%). These patterns of cognitive dysfunction were observed in all four subsets: verbal comprehension, perceptual reasoning, working memory, and processing speed. Thus, the VCI, PRI, WMI, and PSI were below the corresponding mean minus 1 SD values for the healthy controls (<85) in seven (58.3%), six (50.0%), seven (58.3%), and eight (66.7%) patients, respectively.

Table 1. Intellectual function of the EA2 patients on the Wechsler Intelligence Scales.

Patient	Intelligence level	FSIQ	VCI	PRI	WMI	PSI
1	Average	101	90	103	109	110
2	Average	99	100	94	125	81
3	Low average	86	90	103	93	75
4	Low average	83	76	96	81	98
5	Low average	80	77	88	98	81
6*	Low average	80	86	84	83	92
7	Low average	80	83	92	90	78
8	Borderline	77	88	74	75	97
9	Intellectual dysfunction	67	74	68	78	84
10	Intellectual dysfunction	66	70	76	69	84
11	Intellectual dysfunction	64	68	76	78	66
12	Intellectual dysfunction	63	72	72	63	78
Mean±SD		78.8±12.1	81.2±9.4	85.5±11.7	86.8±16.7	85.3±11.5

^*The Korean version of the Wechsler Intelligence Scale for Children—Fourth Edition was used for the intelligence test of Patient 6; the other patients were tested using the Korean version of the Wechsler Adult Intelligence Scale—Fourth Edition.

EA2, episodic ataxia type 2; FSIQ, Full-Scale Intelligence Quotient; PRI, Perceptual Reasoning Index; PSI, Processing Speed Index; SD, standard deviation; VCI, Verbal Comprehension Index; WMI, Working Memory Index.

Brain volumetry

VBM revealed that the GMV was significantly lower in patients with EA2 than in healthy controls with comparable age and sex distributions (Fig. 1 and Table 2). The decreased GMV was mostly localized to the cerebellum, including the vermis (lobules IV, V, and VIII) and bilateral flocculi (lobule X) (p<0.05, FWE-corrected, cluster extent >50 voxels). Brainstem volumetry further demonstrated significant atrophy in EA2 patients (Table 3), with the total volumes of the brainstem as well as of the medulla, pons, and midbrain being significantly reduced relative to the controls (p<0.05, independent-samples t-tests with Welch’s correction). The volume ratio relative to ICV was also decreased for the total brainstem and pons (p<0.05, independent-samples t-tests with Welch’s correction).

Fig. 1. Gray-matter volume was lower in patients with episodic ataxia type 2 than in healthy controls. The results are shown for a significance criterion of p<0.05, with familywise error correction and a cluster extent of >50 voxels. The coordinates are indicated using the Montreal Neurological Institute coordinate system.

Table 2. Voxel-based morphometry results: gray-matter group differences and correlations with intelligence indices in EA2 patients.

Brain region			Side	Peak MNI space (x, y, z)	k	Peak t	p
Group comparisons (EA2<control)
	Cerebellum, vermis VIII (pyramis)		M	1.5, -66.0, -31.5	1,767	8.74	<0.001*
	Cerebellum, vermis IV & V (culmen)		M	1.5, -49.5, -21.0		6.87	0.004*
	Cerebellum, lobule X (flocculus)		R	22.5, -39.0, -45.0	225	7.65	<0.001*
	Cerebellum, lobule X (flocculus)		L	-21.0, -37.5, -43.5	130	7.28	0.001*
Positive correlations with intelligence indices
	FSIQ
		Parahippocampal gyrus	L	-37.5, -12.0, -25.5	56	7.69	<0.001†
		Caudate nucleus	R	19.5, 13.5, 9.0	155	6.48	<0.001†
		Cerebellum, crus II	R	15.0, -76.5, -36.0	65	5.28	<0.001†
	VCI
		Cerebellum, crus II	L	-19.5, -79.5, -33.0	2,465	8.95	<0.001†
		Cerebellum, crus I	L	-36.0, -66.0, -31.5		6.89	<0.001†
		Cerebellum, crus I	R	21.0, -76.5, -30.0	2,732	8.88	<0.001†
		Cerebellum, crus II	R	34.5, -82.5, -37.5		7.37	<0.001†
	PSI
		Parahippocampal gyrus	R	21.0, -24.0, -15.0	65	6.87	<0.001†

^*Voxel-level familywise error-corrected p value; ^†Voxel-level uncorrected p value.

FSIQ, Full-Scale Intelligence Quotient; k, cluster extent (number of contiguous voxels); L, left; M, midline; MNI, Montreal Neurological Institute; Peak t, peak t-statistic from the general linear model at the local maximum (subpeaks within the same cluster were indented and k was not repeated); PSI, Processing Speed Index; R, right; VCI, Verbal Comprehension Index.

Table 3. Brainstem atrophy in the study subjects.

		Healthy controls		EA2 patients		p
		cm3	%	cm3	%	cm3	%
Intracranial volume		1,528.82±183.69		1,501.05±145.61		0.656
Brainstem volume		27.07±2.69	1.79±0.23	24.56±1.36	1.64±0.11	0.003	0.032
	Medulla	4.75±0.50	0.31±0.04	4.30±0.30	0.29±0.03	0.005	0.059
	Pons	15.54±1.80	1.03±0.15	14.03±0.98	0.94±0.06	0.007	0.035
	Midbrain	6.52±0.52	0.43±0.05	6.02±0.31	0.40±0.03	0.003	0.068

Data are mean±standard deviation values.

EA2, episodic ataxia type 2.

Correlations of brain volumes with intelligence

Exploratory voxel-wise analyses revealed various associations between GMV and intelligence indices (Figs. 2, 3, and Table 2). The FSIQ was positively correlated with the GMVs in the left parahippocampal gyrus, right caudate nucleus, and right cerebellar crus II (p<0.001, uncorrected, cluster extent >50 voxels). The VCI also showed a positive correlation with the GMVs in the bilateral cerebellar crura I and II, and the PSI was positively correlated with the GMV in the right parahippocampal gyrus. In contrast, the PRI and WMI were not significantly correlated with GMVs. Fig. 3 presents scatter plots of the relationships between cognitive indices and regional GMVs in the significant clusters.

Fig. 2. Correlations of gray-matter atrophy with the FSIQ (A) and VCI (B) from the Korean version of the Wechsler Adult Intelligence Scale or the Korean version of the Wechsler Intelligence Scale for Children in patients with EA2. The results are shown for a significance criterion of p<0.001 (uncorrected, and cluster extent >50 voxels). Coordinates are indicated using the MNI coordinate system. EA2, episodic ataxia type 2; FSIQ, Full-Scale Intelligence Quotient; MNI, Montreal Neurological Institute; VCI, Verbal Comprehension Index.

Fig. 3. Scatter plots of correlations between cognitive indices and regional GMVs in the significant clusters identified by voxel-based morphometry. FSIQ versus GMVs in the left parahippocampal gyrus (A) (r=0.899), right caudate nucleus (B) (r=0.867), and right cerebellar crus II (C) (r=0.860). VCI versus GMVs in left cerebellar crura I and II (D) (r=0.923) and right cerebellar crura I and II (E) (r=0.960). PSI versus GMV in the right parahippocampal gyrus (F) (r=0.895). The GMVs represent standardized residuals after regressing out age and the total intracranial volume. Pearson correlation coefficients and p values are shown. Note that these correlations are subject to circularity bias since the clusters were defined by the same cognitive variables. FSIQ, Full-Scale Intelligence Quotient; GMV, gray matter volume; PSI, Processing Speed Index; VCI, Verbal Comprehension Index.

DISCUSSION

This study has provided the first comprehensive documentation of brain atrophy in a series of patients with a pure phenotype of EA2. The findings include significant atrophy primarily involving the cerebellum and brainstem, and we also observed positive correlations between the GMVs in specific brain regions and measures of intelligence. Specifically, the FSIQ was positively correlated with the GMVs in the right caudate nucleus, left parahippocampal gyrus, and right cerebellar crus II. Additionally, the VCI was correlated with the GMVs in bilateral cerebellar crura I and II, while PSI was correlated with the GMV in the right parahippocampal gyrus. These exploratory results suggest that there is a complex relationship between brain structures and cognitive function in EA2, thus extending our understanding beyond the traditionally recognized motor symptoms of this disorder.

Cerebellar atrophy, brainstem atrophy, and ocular motor dysfunction in EA2

The VBM analyses revealed that the cerebellar volume was significantly decreased in patients with EA2. Given the mutations found in CACNA1A encoding P/Q-type voltage-gated calcium channels that are strongly expressed in cerebellar Purkinje and granule cells,¹ cerebellar dysfunction and atrophy are to be expected in patients with EA2. These channels participate in the proliferation and differentiation of neurons, outgrowth and synaptogenesis of neurites, excitability of neuronal membranes, signal transduction, gene expression, and neurotransmitter release.¹⁵,¹⁶,¹⁷,¹⁸ Consistent with this molecular basis, the interictal and ictal phenomena in EA2 are predominantly related to cerebellar dysfunction, particularly involving the vestibulocerebellum including the flocculus. While this cerebellar involvement is expected, the present study is the first to perform quantitative volumetric analyses of cerebellar atrophy in EA2. Previous studies have applied qualitative visual analyses to smaller case series,¹⁹,²⁰ while brain MRI findings were reported to be normal in most cases during routine clinical assessments. There is a single case report of widespread cerebellar hypoperfusion in SPECT, but not structural changes.¹⁹ Our study has performed the first quantitative analyses revealing cerebellar atrophy patterns in genetically confirmed EA2 patients, with clear atrophy observed in the brainstem, cerebellar vermis (lobules IV, V, and VIII), and bilateral flocculi.

These affected cerebellar areas are primarily involved in the control of ocular motor and vestibular function.²¹ Indeed, patients with EA2 mostly show interictal ocular motor abnormalities that include downbeat nystagmus, gaze-evoked nystagmus, positional nystagmus, impaired smooth pursuit, and saccadic dysmetria.³ All of these findings manifested with lesions involving the vestibulocerebellum (the nodulus and flocculus) and dorsal ocular motor vermis.²¹ Cerebellar atrophy has also been observed in spinocerebellar ataxia type 6 and familial hemiplegic migraine, which are allelic disorders of EA2.²²,²³,²⁴

Our regional brain volume measurements also revealed brainstem atrophy in the patients with EA2. The brainstem atrophy was a rather unexpected finding in EA2 since brainstem dysfunction is not a prominent feature of this condition.³ However, a recent study found vertical saccadic slowing—indicating midbrain dysfunction—in a patient with EA2.²⁵ In addition, a case of EA2 presenting with torsional saccadic palsy supports brainstem involvement.²⁶ Frequent associations of migraine and weakness during episodes also suggest brainstem dysfunction in EA2.³ It should be noted that multiple comparisons across brainstem substructures were not formally corrected, and so these results should be interpreted with caution.

Neural correlates of intellectual dysfunction in EA2

We recently reported intellectual dysfunction in patients with EA2.⁶ The prevalence of intellectual dysfunction was higher than that found in previous studies involving patients with CACNA1A mutations (50%–61%). However, these estimates should be interpreted with caution due to substantial differences in both the cohort compositions and methodologies in previous studies. Some studies have included a broad spectrum of CACNA1A-related phenotypes beyond EA2,²⁷ with cognitive dysfunction often being determined based on clinical neurodevelopmental assessments rather than standardized intelligence testing.²⁰,²⁷ In contrast, our cohort consisted exclusively of genetically confirmed EA2 patients who underwent standardized intellectual assessments. Differences in phenotypic composition and evaluation methods may therefore at least partially explain the higher prevalence of intellectual dysfunction observed in our study. In addition, our recruitment of patients from three tertiary referral centers specialized in neuro-otology and movement disorders may have preferentially captured more severely affected individuals, which may constitute an inherent selection bias.

This study further found that the GMVs in the cerebellum, parahippocampal gyrus, and caudate nucleus were positively correlated with measures of intelligence. These findings align with and extend the previous neuroimaging evidence linking subcortical volumes with cognitive function in healthy populations. Several studies have found modest but consistent positive correlations between the caudate nucleus volume—particularly for fluid and structural aspects—and intelligence measures in healthy children and young adults.²⁸,²⁹,³⁰,³¹

The involvement of the cerebellum in cognitive function is particularly noteworthy. Since cognitive and limbic functions are also represented by the cerebellum, cerebellar dysfunction may result in deficits of executive function, visual spatial processing, linguistic skills, regulation of affect, and social interaction (cerebellar cognitive affective syndrome [CCAS]).³²,³³,³⁴ The calcium currents that flow through the P/Q-type calcium channel participate in early maturation of the cerebellum, and dysfunction of this channel may give rise to cognitive-affective dysfunction during early life by altering cerebellar tuning of cognitive cortical networks.³⁵

These extracerebellar GM changes may be mechanistically understood within the framework of the CCAS.³⁶,³⁷ The CCAS posits that the cerebellum modulates cognitive and affective processing through dense reciprocal connections with the associative cortical and subcortical regions via the cerebellar–thalamo–cortical (CTC) circuit. The largest clusters of GM correlations in our patients were localized to cerebellar crura I and II, which are recognized as the primary cognitive territories of the cerebellum and are functionally connected to the frontoparietal network and default-mode network.³⁸,³⁹,⁴⁰ Within the CTC circuit, the dentate nucleus projects via the superior cerebellar peduncle to the contralateral ventrolateral thalamus, which in turn sends efferents to the dorsolateral prefrontal cortex and other associative cortical areas.³² Disruption of this circuit by cerebellar pathology—particularly in the cognitive lobules (crura I and II)—may produce secondary transneuronal changes in remote but anatomically connected structures. The right caudate nucleus is a key component of the cortico–striato–thalamo–cortical loop that participates in executive function and learning, while the parahippocampal gyrus mediates episodic memory encoding and contextual processing. These findings extend previous observations of CACNA1A-related neuropsychiatric phenotypes beyond the cerebellum⁵ and align with accumulating evidence that EA2 represents a multisystem disorder affecting distributed cognitive networks.

The correlations of GMVs in the parahippocampal gyrus and caudate nucleus with intellectual dysfunction suggest that maldevelopment of these cortical and subcortical structures can also contribute to cognitive dysfunction. This is supported by animal studies showing that mice with the selective knockout of CACNA1A in the forebrain or cerebellum exhibited marked deficits in memory, spatial learning, and exploratory behaviors.⁴¹,⁴² Furthermore, an autopsy study found cortical microdysgenesis consisting of nodular heterotopia and hypermyelination in the temporal lobe, in addition to a marked loss of Purkinje and granule cells exclusively in the superior cerebellar vermis, in a patient with mild mental retardation and hereditary acetazolamide-responsive paroxysmal ataxia.⁴³ This histopathological evidence supports the notion that structural abnormalities beyond macroscopic atrophy contribute to cognitive dysfunction in EA2.

These findings collectively suggest that the neural correlates of intellectual dysfunction in EA2 involve a network of brain regions, including the cerebellum, caudate nucleus, and parahippocampal gyrus. The relationships between structural changes in these regions and intellectual dysfunction in EA2 patients might reflect the clinical relevance of these brain–behavior relationships, and that cerebellar and basal ganglia integrity are key factors in preserving cognitive function in this disorder. However, the smallness of the sample, the uncorrected statistical criterion, and the testing of multiple cognitive indices without applying formal correction mean that these correlational findings should be interpreted as hypothesis-generating rather than confirmatory.

Limitations and future directions

Several important aspects need to be considered when interpreting the results of this study. The MRI Biomarkers Working Group of the Ataxia Global Initiative recommends a comprehensive approach to neuroimaging in ataxias, including not only morphometric analysis using 3D T1- or T2-weighted images but also diffusion-tensor imaging to reflect microstructural alterations and structural connectivity, as well as resting-state functional MRI to investigate functional connectivity.⁴⁴ Our study was limited to the analysis of 3D T1-weighted images, which might not capture microstructural or functional alterations that could precede any visible atrophy in EA2 patients. This methodological constraint may explain why certain cerebral regions associated with intellectual measurements in our correlation analyses did not show significant atrophy in the VBM group analyses.

Future research should focus on longitudinal studies to track the progression of these structural changes and their relationships with cognitive dysfunction in EA2. Additionally, the implementation of the comprehensive multimodal imaging protocol recommended by the MRI Biomarkers Working Group could provide a more-nuanced understanding of the neural mechanisms underlying both motor and cognitive symptoms in EA2.

Several additional limitations should be acknowledged. First, the wide age range of the participants (10–52 years) may have resulted in heterogeneity in both brain structure and cognitive function; although age was included as a covariate, nonlinear developmental effects might not have been fully captured. Second, the WAIS-IV contains subtests that involve motor speed or visuomotor coordination, which may be confounded by the ataxia and oculomotor dysfunction observed in EA2. CCAS-specific instruments such as the CCAS Scale would be more suitable for isolating cerebellar-specific cognitive dysfunction in future studies.³⁷ Third, cognitive testing was not performed in the control group; although the WAIS-IV provides age-adjusted normative data, direct cognition comparisons between groups were not possible. Fourth, potential confounding variables such as educational background, socioeconomic status, mood disturbances, and medication exposure were not systematically incorporated in the imaging analyses, and their effects cannot be fully excluded. Fifth, the control group was selected to achieve comparable group-level age and sex distributions rather than individual pair matching, and so the analyses were conducted as independent group comparisons with covariate adjustment. Sixth, the robustness of the voxel-based findings may have been affected by the selected cluster-extent criterion (>50 voxels), and sensitivity analyses using alternative criteria were not performed. Together with the modest sample size, these factors mean that all findings should be interpreted as exploratory and hypothesis-generating only.

Despite the interpretational caveats described above, this study has yielded important findings confirming vestibulocerebellar and brainstem atrophy in clinically and genetically well-defined EA2 patients. Moreover, its findings provide meaningful insights into the brain regions associated with cognitive dysfunction in EA2 patients, thus extending our understanding of the disorder beyond its classical motor symptoms. The correlations between structural changes in specific brain regions and intellectual dysfunction in EA2 patients might reflect the clinical relevance of these brain–behavior relationships. These correlations suggest that cerebellar and basal ganglia integrity are key factors in preserving cognitive function in this disorder, which is consistent with recent literature on the roles of these structures in cognition.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.

Supplementary Materials

The online-only Data Supplement is available with this article at https://doi.org/10.3988/jcn.2025.0760.

Supplementary Table 1

Clinical and genetic characteristics of patients