The sample included 9 individuals with SCA6 and 8 healthy controls. Overall, no significant differences were found between groups regarding age or years of education, whereas estimated IQ showed a borderline difference (p = 0.051), with lower scores in the SCA6 group; the corresponding data are presented in Table 1. Within the SCA6 group, exploratory Spearman correlations indicated that the number of failed CCAS-S domains was not significantly associated with age (ρ = 0.383, p = 0.309; 95% CI [− 0.377, 0.835]) or years of education (ρ = −0.433, p = 0.244; 95% CI [− 0.852, 0.324]), suggesting that these demographic variables did not substantially influence the number of failed CCAS-S domains.

Performance on the Hayling Test, measured by total completion time (mean ± SD, seconds), a time-dependent measure sensitive to inhibitory control, was poorer in the SCA6 group (117.60 ± 74.50, SE 26.34) than in controls (45.29 ± 17.67, SE 6.25; mean difference 72.31, 95% CI [9.58, 135.03]; p = 0.01; d = 1.34). Performance on the Stroop Test, assessed by completion time (mean ± SD, seconds), a time-dependent measure sensitive to selective attention, was also poorer in patients (19.82 ± 4.12, SE 1.37) than in controls (14.30 ± 2.95, SE 1.04; mean difference 5.52, 95% CI [1.83, 9.21]; p = 0.007; d = 1.52). The contribution of motor slowing, reduced processing speed, and verbal output demands to performance on these time-dependent measures cannot be ruled out. In contrast, cognitive flexibility on the Hayling Test (mean ± SD; SCA6: 5.91 ± 3.05, SE 1.08; controls: 4.00 ± 1.38, SE 0.49; mean difference 1.90, 95% CI [− 0.74, 4.54]; p = 0.13; d = 0.80) and executive planning on the Tower of London (total score, mean ± SD; SCA6: 32.80 ± 1.79, SE 0.80; controls: 28.00 ± 7.12, SE 2.69; mean difference 4.80, 95% CI [− 1.83, 11.43]; p = 0.17; d = 0.85) did not differ significantly between groups.

Immediate visuospatial memory, assessed with the Corsi Blocks Forward (maximum span, mean ± SD), was slightly lower in the SCA6 group (7.67 ± 2.29, SE 0.76) than in controls (9.00 ± 1.60, SE 0.57; mean difference − 1.33, 95% CI [− 3.37, 0.70]; p = 0.190; d = − 0.67), although this difference was not statistically significant. Visuospatial working memory, measured with the Corsi Blocks Backward (maximum span, mean ± SD), followed a similar trend (SCA6: 6.67 ± 1.94, SE 0.65; controls: 8.38 ± 2.62, SE 0.93; mean difference − 1.71, 95% CI [− 4.15, 0.73]; p = 0.144; d = − 0.75), without reaching statistical significance. Verbal episodic memory, evaluated with the RAVLT delayed recall score (mean ± SD), was significantly reduced in SCA6 patients (6.44 ± 2.40, SE 0.80) compared with controls (12.00 ± 2.39, SE 0.85; mean difference − 5.56, 95% CI [− 8.04, − 3.07]; p < 0.001; d = − 2.32). Visuoconstructive and visual episodic memory abilities, assessed with the Rey–Osterrieth Complex Figure (copy and delayed recall scores, mean ± SD; maximum = 36 points), a task involving visuoconstructive and visuomotor integration, graphomotor skills, and organizational planning, showed trends toward poorer performance in the SCA6 group both on the copy condition (SCA6: 29.83 ± 8.63, SE 2.88; controls: 36.00 ± 0.00, SE 0.00; mean difference − 6.17, 95% CI [− 12.80, 0.47]; p = 0.062; d = − 0.98) and delayed recall (SCA6: 14.28 ± 8.95, SE 2.98; controls: 20.81 ± 6.39, SE 2.26; mean difference − 6.53, 95% CI [− 14.54, 1.47]; p = 0.108; d = − 0.83), although these differences did not reach statistical significance. The contribution of motor and visuomotor factors to performance on this task cannot be ruled out.

Within the language domain, confrontation naming, assessed with the Boston Naming Test (total correct responses, mean ± SD; maximum score = 60), showed a trend toward poorer performance in the SCA6 group (52.11 ± 6.47, SE 2.16) compared with controls (56.50 ± 2.78, SE 0.98; mean difference − 4.39, 95% CI [− 9.60, 0.82]; p = 0.096; d = − 0.86), although this difference was not statistically significant.

Performance on the WASI Block Design subtest (total raw score, mean ± SD), a task involving visuoconstructive and visuomotor integration, constructional praxis, and manual dexterity under timed administration, was poorer in patients with SCA6 (27.4 ± 13.42, SE 4.74) than in controls (46.4 ± 9.93, SE 3.51; mean difference − 19.00, 95% CI [− 31.76, − 6.24]; p = 0.006; d = − 1.61). The contribution of motor and visuomotor factors to this difference cannot be ruled out. In contrast, performance on WASI Matrix Reasoning (total raw score, mean ± SD; SCA6: 21.9 ± 7.70, SE 2.57; controls: 26.5 ± 4.81, SE 1.70; mean difference − 4.61, 95% CI [− 11.24, 2.02]; p = 0.16; d = − 0.71) did not differ significantly between groups.

Anxiety symptoms, assessed with the HAM-A, were significantly higher in the SCA6 group (mean 11.00 ± 4.47, SE 1.49) than in controls (mean 4.63 ± 3.02, SE 1.07; mean difference 6.38, 95% CI [2.44, 10.31]; p = 0.004; d = 1.65). For reference, the conventional HAM-A severity thresholds are: 0–7, no anxiety; 8–14, mild anxiety; 15–23, moderate anxiety; ≥24, severe anxiety. Depressive symptoms, assessed with the BDI-II, were higher in the SCA6 group (mean 8.56 ± 7.16, SE 2.39) than in controls (mean 4.75 ± 2.66, SE 0.94; mean difference 3.81, 95% CI [− 1.88, 9.49]; p = 0.178; d = 0.69), although this difference did not reach statistical significance. The number of failed CCAS-S domains was higher in the SCA6 group (mean 4.67 ± 2.00, SE 0.67) compared with controls (mean 1.75 ± 0.46, SE 0.16; mean difference 2.92, 95% CI [1.36, 4.47]; p = 0.001; d = 1.95), with a higher number of failed domains and a shift toward more severe categories. Figure 1 illustrates these results.

Clinical and emotional scales by group. Note. Boxplots display median, interquartile range, and individual data points for the Beck Depression Inventory-II (BDI-II), Hamilton Anxiety Scale (HAM-A), and the number of failed CCAS-S domains. Orange boxes represent the SCA6 group and cyan boxes represent the control group. Asterisks indicate statistically significant between-group differences (* p < 0.05)

In accordance with the original CCAS-S criteria, no control participant met criteria for definite CCAS. Among SCA6 patients, 77.8% were classified as definite CCAS, 11.1% as probable, and 11.1% as possible. In the control group, 75.0% were classified as probable and 25.0% as possible CCAS. Table 2 presents the CCAS-S classification in each group.

In the SCA6 group, exploratory correlation analyses revealed significant associations between the number of failed CCAS-S domains and performance on the Boston Naming Test (r = − 0.721, p = 0.028; 95% CI [− 0.937, − 0.110]), Block Design (r = − 0.758, p = 0.029; 95% CI [− 0.953, − 0.114]), estimated IQ (r = − 0.698, p = 0.037; 95% CI [− 0.931, − 0.062]), Stroop Test (ρ = 0.758, p = 0.018; 95% CI [0.188, 0.946]), and Rey Figure Copy (ρ = −0.704, p = 0.034; 95% CI [− 0.932, − 0.076]).

Additional non-significant trends were observed for Corsi Blocks Backward (r = − 0.645, p = 0.060; 95% CI [− 0.917, 0.033]), Matrix Reasoning (r = − 0.652, p = 0.057; 95% CI [− 0.918, 0.022]), Picture Completion (r = − 0.610, p = 0.081; 95% CI [− 0.907, 0.091]), Rey Figure Memory (r = − 0.340, p = 0.371; 95% CI [− 0.819, 0.419]), RAVLT (r = − 0.329, p = 0.387; 95% CI [− 0.815, 0.428]), and HAM-A (r = − 0.531, p = 0.141; 95% CI [− 0.884, 0.206]). No relevant associations were identified for Hayling Test, cognitive flexibility (r = − 0.199, p = 0.637; 95% CI [− 0.792, 0.588]), inhibitory control (r = − 0.159, p = 0.707; 95% CI [− 0.777, 0.615]), Corsi Blocks Forward (r = − 0.409, p = 0.274; 95% CI [− 0.844, 0.350]), BDI-II (r = 0.329, p = 0.388; 95% CI [− 0.429, 0.815]), or Tower of London (r = − 0.722, p = 0.168; 95% CI [− 0.980, 0.442]). Given the small sample size (n = 9 for most analyses; n = 8 for Block Design, due to missing data), all estimates are sensitive to individual observations and should be interpreted with caution (Fig. 2).

Correlations between CCAS-S and cognitive performance in SCA6. Note. Scatterplots showing relationships between the number of failed CCAS-S domains and neuropsychological measures in the SCA6 group (n = 9 for most analyses; n = 8 for Block Design, due to missing data). Pearson correlation coefficients (r) were applied for variables meeting normality assumptions (Shapiro–Wilk p > 0.05); Spearman's ρ was used for Stroop Test and Rey Figure Copy, which did not meet normality criteria. Dashed lines represent linear fit with 95% confidence intervals. Asterisks indicate statistically significant correlations (* p < 0.05). All analyses are exploratory

Both simple and partial correlations were computed between SARA scores and the number of failed CCAS-S domains. The simple correlation yielded ρ = 0.903 (p = 0.002; 95% CI [0.544, 0.982]; n = 8), indicating that greater motor severity was associated with a higher number of failed CCAS-S domains. The partial correlation controlling for disease duration remained high and statistically significant (ρ = 0.849, p = 0.016; 95% CI [0.27, 0.98]), suggesting that the association between motor severity and cognitive-affective impairment persists independently of disease duration. Figure 3 graphically illustrates the exploratory correlation results.

Disease severity and the number of failed CCAS-S domains in SCA6. Note. Scatterplot showing the relationship between SARA scores and the number of failed CCAS-S domains in SCA6 patients (n = 8). Each point represents an individual participant, colored by disease duration. Dashed line represents linear fit with 95% confidence interval (shaded area). Simple and partial correlations were computed using Spearman's ρ. Simple correlation: ρ = 0.903, p = 0.002; 95% CI [0.544, 0.982]. Partial correlation controlling for disease duration: ρ = 0.849, p = 0.016; 95% CI [0.27, 0.98]. All analyses are exploratory