Cognitive impairment and its neuroimaging correlates in spinocerebellar ataxia 2

Abstract

Introduction

Cognitive impairment (CI) is reported but is poorly explored in spinocerebellar ataxia 2 (SCA2). This study was undertaken to evaluate and classify cognitive impairment in patients with SCA2 and to identify their grey matter (GM) correlates.

Methods

We evaluated the neurocognitive profile of 35 SCA2 and 30 age-, gender- and education-matched healthy controls using tests for attention, executive functions, learning and memory, language and fluency, and visuomotor constructive ability. Patients were classified into SCA2 with and without CI based on normative data from population and healthy controls. Furthermore, patients with CI were sub-classified based on the number of impaired domains into multi-domain CI (≥3 domains; MDCI) and limited domain CI (≤2 domains; LDCI). The underlying GM changes were identified using voxel based morphometry.

Results

The mean age at onset, duration of disease, and ataxia score was 28.7 ± 8.51 years, 66.7 ± 44.1 months, and 16.1 ± 4.9 points, respectively. CI was present in 71.4% of SCA2 subjects (MDCI: 42.7%; LDCI: 28.5%). Patients with CI had significant atrophy of the posterior cerebellum, sensorimotor cortex, and superior frontal gyrus (FWE p-value <0.05). Patients with MDCI had significant GM atrophy of the angular gyrus compared to LDCI (FWE p-value <0.05).

Conclusion

Patients with CI had significant GM involvement of the posterior cerebellum and frontal lobe, suggestive of impairment in the cerebello-fronto-cortical circuitry.

1. Introduction

Spinocerebellar ataxia 2 (SCA2) is a rare neurodegenerative disease caused by triplet nucleotides (cytosine-adenine-guanine) repeat expansion in the ATXN2 (12q24.1) gene [1]. The clinical features of SCA2 are highly variable and include cerebellar involvement, oculomotor abnormalities, peripheral neuropathy, movement disorders, motor-neuron signs, pyramidal involvement, autonomic dysfunction, and cognitive dysfunction [2,3]. Furthermore, a recent study by Pedroso et al. also demonstrated high prevalence of REM sleep behaviour disorder, excessive daytime sleepiness, restless leg syndrome, weight loss, and cramps in SCA2 [4].

Previous studies have reported cognitive impairment (CI) in 5–19% of patients with SCA2 [5–7]. However, a meta-analysis by Lindey and Storey [8] pointed out that the low prevalence of CI in SCA2 (SCA2-CI) is due to the lack of comprehensive assessment of neuropsychological domains. The CI in SCA2 ranges from normal cognition to profound CI [8,9]. Compared to SCA 1, 3, 6, and 7, SCA2 is most commonly associated with dementia [8]. The affected cognitive domains in SCA2-CI shows significant variation [6–9]. While executive dysfunction and deficits in learning and memory are consistently reported, patients with profound CI may involve other domains [6–9]. However, the extent of involvement of other cognitive domains are variable and without clear consensus.

Cognitive dysfunction is well studied in chronic degenerative diseases such as Alzheimer’s dementia and Parkinson’s disease [10,11]. In both these disorders, CI is classified based on the number of domains affected and their functional consequence [10,11]. Although CI in SCA2 is well known, the severity of involvement, progression of CI, inter-individual differences, and their neural underpinnings are not well understood [10,11]. Despite severe cerebellar atrophy, previous neuropsychological studies have reported a predominant frontal-subcortical type of impairment in SCA2-CI [6–9,12,13]. Concurrent evaluation of patients with SCA2 using comprehensive cognitive assessment and neuroimaging may shed light on the pathophysiological substrates underlying SCA2-CI.

In this study, we evaluated patients with SCA2 using comprehensive neuropsychological evaluation and classified individual patients using normative data from age, gender, education matched healthy population. We classified patients based on their performance on the neuropsychological tests and then applied voxel-based morphometry (VBM) to identify the grey matter (GM) changes underlying CI in SCA2.

2. Methods

This study was performed at the National Institute of Mental Health and Neurosciences (NIMHANS), India. We recruited 35 patients with genetically proven SCA2 and 30 healthy controls (HC) after obtaining ethical approval. All subjects provided informed consent. The inclusion criteria for patients were (i) genetic diagnosis of SCA2, (ii) age more than 18 years, (iii) absence of other neurological comorbidities, and (iv) formal school education. The healthy controls did not have any neurological comorbidities and were age, gender and education matched. Patients (n = 23) and HC (n = 20) from this study have previously been used in a previous study from our group [14].

2.1. Clinical and demographic details

All subjects underwent neurological assessment using the Scale for the Assessment and Rating of Ataxia (SARA), Cognitive functional rating scale (CFRS), Nine-hole peg test of dominant hand (9HPT), PATA speech test (PST), Neuropsychiatric inventory (NPI), and Hamilton’s scale for anxiety and depression (HAM-A, HAM-D). The ‘age at onset (AAO)’, ‘duration of disease (DOD)’ and CAG repeat size was also obtained.

2.2. Assessment of cognitive impairment

We used ten validated tests to assess the cognitive domains. These included (a) colour trails-1 test (CT1) and serial subtraction test (SST) for focussed attention; (b) colour trails-2 (CT2) and Stroop effect for executive function; (c) delayed recall scores of complex figure test (CFT-DR) and auditory verbal learning test (AVLT-DR) for learning and memory; (d) controlled oral word associate test (COWA) and animal naming test (ANT) for language and fluency; and (e) copy scores of complex figure test (CFT-copy) and pentagon drawing test (PDT) for visuomotor construction function (Supplementary Tables 1 and 2) [15–17].

We used normative data for Indian population to score each subjects’ neuropsychological performance [15]. A score less than two standard deviation of the normative data was the threshold for impairment in a particular domain [15] (see Supplementary Table 1 for cut-off scores).

2.3. Classification of cognitive impairment

Parkinson’s disease and SCA2 are believed to have a fronto-striatal origin of CI [8]. Hence, we used the criteria for CI in Parkinson’s disease to classify patients with SCA2 in this study [18,19]. Firstly, we classified patients into SCA2 with CI (SCA2-CI) and SCA2 with no CI (SCA2-NCI) on the basis of presence or absence of impaired score on neuropsychological tests. Patients with SCA2-CI were further sub-classified into multidomain CI (MDCI) and limited domain CI (LDCI). LDCI consisted of patients with two impaired test scores in the same cognitive domain or at least one impaired test score in two different cognitive domains. The presence of impaired scores in more than two domains signified MDCI [18,19].

2.4. Neuroimaging in spinocerebellar ataxia 2

All subjects underwent MRI in a Siemens SKYRA 3-T MRI system. A high-resolution, 3D T1 weighted Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence covering the whole brain with TR = 2300 ms, TE = 2.98 ms, flip angle = 9°, sense-factor = 3.5, FOV = 256 × 256 × 160 mm, voxel size = 1 × 1 × 1 mm, slice thickness = 1 mm, acquisition matrix = 256 × 256, and total scan time = 9 min 50 s with 192 slices, was performed.

We used Statistical parametric mapping 12 (SPM12) software, Computational anatomical toolbox 12 (CAT12), and MATLAB R2015a for VBM analysis to identify GM areas with significant difference in local concentrations (density). After conversion of the raw T1-MPRAGE sequences from Digital imaging and communications in medicine (DICOM) format to SPM compatible Neuroimaging informatics technology initiative (NIfTI) format, the images were manually reoriented to the anterior commissure-posterior commissure (AC-PC) plane in SPM12. The T1 weighted anatomical images were spatially normalized by reorienting the images into the same stereotactic space and then segregated into GM, WM, and CSF space, followed by modulation and smoothening with an 8-mm-full-width-half-maximum Gaussian kernel. The smoothened images underwent statistical analysis using general linear modelling. We used unpaired t-test with age, gender and total intracranial volume (TIV) as covariates followed by correction for multiple comparison using the theory of Gaussian random fields (p < 0.05, FWE corrected) to identify changes in GM concentrations among different group. The output of the process is a statistical parametric map showing areas with significant alterations of GM density/concentration. We identified these areas of significant GM loss in Montreal neurological institute (MNI) coordinates and created representative images using the XJview software.

2.5. Statistical analysis

For comparison of variables with normal distribution, we used t-test or ANOVA with post-hoc analysis. Nominal variables were analysed using chi square test and Fisher’s exact test. Multivariate analyses was performed using multinomial logistic regression to identify significant predictive factors for CI. All p-values were interpreted for significance after correcting for multiple comparisons.

3. Results

We recruited 35 patients with SCA2 and 30 healthy controls. There was no significant difference between the mean age, gender distribution or education among the groups. The demographic and clinical variables for the overall SCA2 group is provided in Tables 1 and 2.

Table 1. Clinical and demographic characteristics of SCA2 and subgroups.

Demographic parameters	Patients				Control	P- value
	SCA2 (Overall)	MDCI	LDCI	SCA2- NCI		P1/P2
Number of subjects (families)	35 (29)	15 (15)	10 (6)	10 (8)	30 (30)	–
Gender distribution (Men:Women)	22:13	10:5	7:3	5:5	21:9	0.65/0.35
Education (HE: LE)	23:12	12:3	7:3	4:6	24:6	0.23/0.41
Mean age (in years)	34.0 ± 10.0	36.6 ± 10.4	29.0 ± 6.5	32.8 ± 9.78	36.1 ± 11.88	0.45/0.20
Mean age at onset (in years)	28.7 ± 8.51	29.1 ± 9.0	28.2 ± 8.4	27.6 ± 7.89	–	-/0.90
Mean duration of disease (in months)	66.7 ± 44.1	64.2 ± 48.0	60.0 ± 25.9	62.4 ± 51.1	–	-/0.97
Mean CAG repeats	42.5 ± 4.6	42.4 ± 5.0	42.8 ± 4.18	42.0 ± 2.0	22.1 ± 0.3	-/0.92
Mean SARA score	16.1 ± 4.9	16.7 ± 5.54	15.4 ± 1.90	14.5 ± 5.76	0.1 ± 0.10	0.02*/0.55
Mean 9HPT-dom (in sec)	50.0 ± 16.2	48.2 ± 14.7	51.4 ± 12.2	50.7 ± 19.4	13.5 ± 4.8	0.02*/0.86
Mean PST score (in sec)	29.0 ± 7.18	27.7 ± 6.8	30.8 ± 7.2	29.6 ± 5.64	25.7 ± 9.0	1.0/0.54
Mean CFRS score	6.2 ± 4.26	8.3 ± 2.11	6.6 ± 7.43	4.0 ± 6.39	4.2 ± 7.18	0.63/0.35

P1: p-value of SCA2 vs HC (t-test); P2: p-value of SCA2-MDCI vs SCA2-LDCI vs SCA2-NCI (ANOVA); Significant p-values are denoted by * (after post-hoc analysis). HE: Formal school education ≥10 years; LE: Formal school education less than 10 years; SARA: Scale for assessment and rating of ataxia; 9HPT-dom: nine-hole peg test score of dominant hand; PST: PATA speech test score; CFRS: Cognitive functional rating score.

Table 2. Psychiatric assessment in SCA2.

Neuropsychiatric involvement	SCA2 (Overall)	MDCI	LDCI	SCA2- NCI	P value
Number of subjects	35	15	10	10	–
HAM-A score	8.03 ±	7.88 ±	7.71 ±	8.62 ±	0.95
	6.82	4.25	2.98	12.51
HAM-D score	6.65 ±	6.23 ±	5.0 ±	9.0 ±	0.83
	6.49	3.78	2.58	11.79
NPI domains involved	65.7% (23)	86.6%	60%	40% (4)	0.04*
		(13)	(6)
Delusion	0	0		0	0-
Hallucination	0	0	0	0	–
Agitation	11.4% (4)	6.6% (1)	10%	20% (2)	0.39
			(1)
Depression	31.4% (11)	33.3%	40%	20% (2)	0.34
		(5)	(4)
Anxiety	51.4% (18)	73.3%	50%	20% (2)	0.11
		(11)	(5)
Apathy	2.8% (1)	0	0	10% (1)	0.21
Irritability	5.7% (2)	6.6% (1)	0	10% (1)	0.60
Euphoria	5.7% (2)	6.6% (1)	0	10% (1)	0.60
Disinhibition	20% (7)	13.3%	30%	20% (2)	0.23
		(2)	(3)
Aberrant behaviour	0	0	0	0	–

^*significant p value. HAM-A and HAM-D scores are represented as mean ± standard deviation; Impairment on NPI scale and its subdomains are given as prevalence- % (number). SCA2: Spinocerebellar ataxia 2; MDCI: Multidomain cognitive impairment; LDCI: Limited domain cognitive impairment; SCA2-NCI: No cognitive impairment; NPI: Neuropsychiatric inventory; HAM-A: Hamilton’s scale for anxiety; HAM-D: Hamilton’s scale for depression.

3.1. SCA2 with and without cognitive impairment (MDCI, LDCI, and SCA2-NCI)

Patients with MDCI, LDCI and SCA2-NCI constituted 42.7% (n = 15), 28.5% (n = 10), and 28.5% (n = 10) of the overall SCA2 cohort, respectively. Only 15.1% (n = 5; all MDCI) of patients reported CI affecting their in daily activities as a presenting complaint. There was no significant difference between the three groups with respect to demographic or clinical variables (Table 1). Impairment on NPI scale was significantly more prevalent in MDCI (86.6%) compared to LDCI (60%) and SCA2-NCI (40%) (Fisher’s exact test p = 0.04) (Table 2).

3.2. Cognitive domain impairments in SCA2

Patients with SCA2-CI had impairment in visuomotor construction ability (80%; n = 20), language and fluency (76%; n = 19), learning and memory (60%; n = 15), focussed attention (60%; n = 15), and executive function (24%; n = 6). Patients with SCA2-NCI and healthy controls did not have any impairment in the cognitive domains.

The most prevalent domain impairment in MDCI was impairment in visuomotor construction (100%; n = 15) and focussed attention (100%; n = 15), followed by impairment in learning and memory (86.6%; n = 13), language and fluency (80%; n = 12), and executive dysfunction (26.6%; n = 4).

In patients with LDCI, the most common cognitive domain affected was language and fluency domain (70%; n = 7) followed by visuomotor construction domain (50%; n = 5), learning and memory domain (20%; n = 2), executive dysfunction (20%; n = 2). Patients with LDCI did not have impairment in focussed attention.

The MDCI group had significantly higher proportion of patients with impaired scores in focussed attention (100% vs 0%; Fisher’s exact test p = 0.02), visuomotor construction (100% vs 50%; Fisher’s exact test p = 0.02), and learning and memory (86.6% vs 20%; Fisher’s exact test p = 0.01) domains. The SCA2-LDCI group did not have higher proportion of patients with impaired scores of any domains compared to MDCI.

3.3. Predictors for cognitive impairment in SCA2

Multivariate logistic regression using classification into MDCI, LDCI, and SCA2-NCI as dependent variable and age, gender, education, DOD, SARA, and CAG size as independent variables revealed ‘formal education ≥10 years’ to decrease the multinomial log-odds of being classified into MDCI group by –3.125 units when other independent variables are kept constant (exp. odds: –3.125. confidence interval: 0.004–0.473, p = 0.01). Other variables entered into the model were not statistically significant.

3.4. Neuroimaging in SCA2

Patients with SCA2-CI had significant GM atrophy of the cerebellum (bilateral H–V and H-VI, left H-IX), bilateral sensorimotor cortex, and left superior frontal gyrus (FWE corrected p < 0.05) compared to SCA2-NCI (Table 3, Fig. 1). There were no areas of significant GM atrophy in SCA2-NCI compared to SCA2-CI.

Table 3. Areas with significant grey matter atrophy in (a) SCA2 with and without cognitive impairment (SCA2-CI vs SCA2-NCI) (b) SCA2 with multi-domain and limited domain cognitive impairment (MDCI vs LDCI).

X	Y	Z	Peak T	P value (FWE)	Structures
Areas showing significant grey matter atrophy in SCA2-CI compared to SCA2-NCI
–18.92	–51.29	–22.52	16.9	0.00	Left cerebellum H–V
– 34.04	– 55.62	– 35.35	16.6	0.00	Left cerebellum H-VI
– 20.36	– 61.32	– 20.79	6.50	0.00	Left cerebellum H-IX
22.74	– 59.63	– 25.31	15.8	0.00	Right cerebellum H–V
19.88	– 57.47	–18.40	7.80	0.00	Right cerebellum H-VI
20.31	– 31.65	63.78	6.6	0.00	Right sensorimotor cortex
– 41.82	– 23.38	39.19	6.7	0.00	Left sensorimotor cortex
– 27.38	43.7	16.06	6.3	0.00	Left superior frontal gyrus
Areas showing significant grey matter atrophy in MDCI compared to LDCI
–51.01	–54.03	19.5	8.31	0.02	Left angular gyrus

SCA2: Spinocerebellar ataxia 2; CI: cognitive impairment; NCI: no cognitive impairment.

Fig. 1.

Areas showing significant grey matter atrophy in (a) SCA2 with and without cognitive impairment-cerebellum (bilateral H–V and H-VI, left H-IX), bilateral sensorimotor cortex, and left superior frontal gyrus (b) MDCI versus LDCI - left angular gyrus.

Patients with MDCI had significantly more GM atrophy of the left angular gyrus (FWE corrected p < 0.05) when compared to LDCI and SCA2-NCI groups (Table 3, Fig. 1). The LDCI and SCA2-NCI groups did not have significant GM atrophy when compared to MDCI.

3.5. Clinical correlates of GM atrophy in SCA2 subgroups

Patients with SCA2-CI demonstrated correlation with a trend to significance between (i) GM atrophy of bilateral cerebellum (H-VI) and SARA (p = 0.060). There was no correlation which were significant or had a trend to significance between GM atrophy of any region with DOD, AAO, CAG size, or CFRS score.

Patients with MDCI had significant correlation between GM atrophy of left superior frontal gyrus and worsening CFRS score (FWE corrected p < 0.05) as well as with DOD (FWE corrected. p < 0.05). There was no significant correlation between GM atrophy with AAO, SARA score, and CAG size. The LDCI and SCA2-NCI groups did not show any significant GM areas with clinical or genetic correlations.

4. Discussion

This study identified a higher prevalence of CI in SCA2 (71.4%) in spite of short duration of ataxic symptoms (mean ~5.5 years). Three cognitive severity levels were identified- MDCI (42.7%), LDCI (28.5%), and SCA-NCI (28.5%). SCA-CI (MDCI + LDCI) had significant atrophy of the posterior cerebellum, sensorimotor cortex, and superior frontal gyrus compared to SCA2-NCI. Patients with MDCI had significantly higher GM atrophy of angular gyrus compared to LDCI.

Only few studies have comprehensively assessed the cognitive profile of SCA2. A study by Burk et al. evaluated 14 patients of SCA2 and identified impairment restricted to learning and memory, and executive function [7]. In agreement, Fancellu et al. and Le Pira et al. also reported executive dysfunction and impairment in learning and memory in addition to attention deficits [20,21]. The presence of language and visuospatial dysfunction was reported only in a study by Orsi et al. [22]. In contrast, this study identified impairment in SCA2 to involve all five key cognitive domains-attention, learning and memory, language and fluency, executive function, and visuospatial domains. However, not all subjects had impairment in all domains. It is interesting to note that MDCI had higher proportion of impaired scores in learning and memory, focussed attention, and visuospatial construction domains despite similar demographic, clinical and genetic profile compared to LDCI. Hence, it is possible that MDCI and LDCI may be akin to cognitive severity states such as dementia and mild cognitive impairment (MCI) in Alzheimer’s disease and Parkinson’s disease. Furthermore, patients with LDCI may worsen and evolve into MDCI over time. This is also in agreement with the natural history of SCA2 where mild cognitive dysfunction worsens progressively to dementia in the severe ataxia stage. Previous longitudinal studies have also provided evidence for worsening of specific cognitive dysfunction over time [20,21]. In this study, the correlation obtained between GM atrophy with increasing DOD in MDCI and with SARA score in SCA2-CI (at a reduced threshold of p = 0.06) may also indirectly points towards the same. However, regression analysis did not find DOD or SARA score to predict MDCI, LDCI or SCA2-NCI. These antithetical observations may be due to the relatively small cohort size, non-linear progression of severity of ataxia, recollection bias, and short mean DOD and SARA score (mean DOD: ~5.5 years, mean SARA: ~16.1 points) in our cohort suggestive of early disease. Given the cross-sectional design of our study, it is not possible to gain further insight with regard to conversion from LDCI to MDCI. Another valid result from our study was that higher education was found to decrease the odds of developing MDCI. This is probably because a higher cognitive reserve may prevent worsening of cognitive functions. This observation is also reported in other dementing illnesses [23].

The pathophysiological basis for CI in SCA2 is postulated to be related to (i) dysmetria of thought/cerebellar cognitive affective syndrome (CCAS) or (ii) fronto-striatal dysfunction. The ‘dysmetria of thought’ theory implicates cerebellum as a critical modulator of cognitive, sensorimotor, and limbic functions and may cause impairment in executive functions, visuomotor construction dysfunction, expressive language dysfunction, learning and memory, and attention [24]. Furthermore, the impairments in CCAS are known to be subclinical without significant effect on the activities of daily living and is demonstrable only on detailed neuropsychological assessment [24]. On the other hand, the fronto-striatal circuit (comprised of basal ganglia-thalamus-prefrontal cortex) dysfunction causes a ‘fronto--subcortical’ type of CI with predominant executive dysfunction impairing day to day activities [9]. The presence of subtle neuropsychological impairment in the setting of significant GM atrophy of the posterior cerebellum along with the spectrum of neuropsychiatric features in our cohort supports the possibility of CCAS as an underlying pathogenic mechanism in SCA2. This is also supported by the observations of Olivito and colleagues [25,26].

Although the involvement of the superior frontal gyrus in this study may implicate fronto-striatal dysfunction, the lack of significant atrophy of the thalamus and basal ganglia in SCA2-CI may prove antithetical. However, given that a minority of our patients from MDCI had clinically symptomatic cognitive dysfunction with impairment of ADL, and significant correlation between GM of the superior frontal gyrus with worsening CFRS score and increasing DOD, a partial but significant contribution from the cerebello-frontal cortical loops should be considered, at least when multiple cognitive domains are involved. The superior frontal gyrus being richly connected to the anterior and mid-cingulate cortices (critical nodes of the cognitive control network and default mode network), middle and inferior frontal gyri (involved in the cognitive execution network), sensorimotor areas, thalamus, caudate, and frontal operculum (nodes of the sensorimotor control network) may cause cognitive dysfunction reported in our study [27]. The sensorimotor cortex is traditionally regarded as motor-sensory area. However, previous fMRI studies shows their activation during cognitive tasks not involving motor activity such as motor imagery, mental rotation, social cognition, working memory, language, and auditory processing [28]. The involvement of angular gyrus in MDCI compared to LDCI may also have significant cognitive outcomes. Due to its location at the junction of the temporal, occipital, and parietal lobes, the angular gyrus is an interface that conveys and integrates information from many subsystems. It is known to subserve several cognitive functions such as semantic processing, comprehension, memory retrieval, attention and visuospatial cognition, social cognition and is also a crucial area of the default mode network [29]. Angular gyrus atrophy in patients with MDCI may hence predispose them to develop more deficits in visuo-motor construction, learning and memory, and attention compared to LDCI. Hence, the regional GM atrophy in SCA2-CI can also explain the overall results of the neuropsychological evaluation.

This study had few strengths and limitations. We classified individual subjects based on the domains affected using normative data from the age-, gender-, and education-matched population. This led to identification of severity levels ranging from absent CI to LDCI and MDCI in SCA2 with distinct GM involvement. Previous studies have relied on measures of central tendency such as mean and median of the raw scores in a group wise analysis. These measures are affected by the skewness and presence of outliers, especially when significant inter-individual variability is expected. Secondly, the sample size is relatively larger compared to previous neuroimaging and neuropsychological studies in SCA2.

A major limitation of this study is the cross-sectional design which provides only limited clarity on the possible worsening of cognitive function across time. It should also be borne in mind that despite impairment in multiple cognitive domains, most patients have only subtle deficits and may not complain of ‘dementia’ or functional disability in clinical practice. However, a minority of pedigrees with higher rates of dementia have been reported [8,9]. This may implicate other familial factors than those assessed in this study. Nonetheless, cognitive dysfunction should be assessed routinely and early cognitive rehabilitation attempted.

To conclude, cognitive dysfunction may be present on neuropsychological assessment even in the early stages of SCA2. However, due to the limitation of the study, our findings are to be regarded as initial observations and has to be carefully interpreted and confirmed in longitudinal studies.

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.parkreldis.2021.02.028.