Visuomotor integration deficits are common to familial and sporadic preclinical Alzheimer’s disease

Abstract We investigated whether subtle visuomotor deficits were detectable in familial and sporadic preclinical Alzheimer’s disease. A circle-tracing task—with direct and indirect visual feedback, and dual-task subtraction—was completed by 31 individuals at 50% risk of familial Alzheimer’s disease (19 presymptomatic mutation carriers; 12 non-carriers) and 390 cognitively normal older adults (members of the British 1946 Birth Cohort, all born during the same week; age range at assessment = 69–71 years), who also underwent β-amyloid-PET/MRI to derive amyloid status (positive/negative), whole-brain volume and white matter hyperintensity volume. We compared preclinical Alzheimer’s groups against controls cross-sectionally (mutation carriers versus non-carriers; amyloid-positive versus amyloid-negative) on speed and accuracy of circle-tracing and subtraction. Mutation carriers (mean 7 years before expected onset) and amyloid-positive older adults traced disproportionately less accurately than controls when visual feedback was indirect, and were slower at dual-task subtraction. In the older adults, the same pattern of associations was found when considering amyloid burden as a continuous variable (Standardized Uptake Value Ratio). The effect of amyloid was independent of white matter hyperintensity and brain volumes, which themselves were associated with different aspects of performance: greater white matter hyperintensity volume was also associated with disproportionately poorer tracing accuracy when visual feedback was indirect, whereas larger brain volume was associated with faster tracing and faster subtraction. Mutation carriers also showed evidence of poorer tracing accuracy when visual feedback was direct. This study provides the first evidence of visuomotor integration deficits common to familial and sporadic preclinical Alzheimer’s disease, which may precede the onset of clinical symptoms by several years.


Dual-task cost in FAD
As the FAD participants completed both dual-and single-task circle-tracing trials (with and without concurrent serial subtraction), their tracing performance can be compared between the two. Results presented in the main manuscript show that tracing was slower in the dual-task, but more accurate (Table 1, Table 2). The improvement in accuracy can be explained by speed-accuracy trade-offs: as the two tasks competed for cognitive resources, participants were obliged to slow down, limiting the likelihood of adopting a hasty (i.e. fast but inaccurate) tracing style.
The reduction in speed can be quantified for each individual as a 'dual-task cost', which is a measure of the extent to which tracing speed was compromised by having to allocate cognitive resources to the subtraction task. Dual-task cost was calculated from the mean number of rotations completed in singleand dual-task trials (combined across the direct and indirect conditions) as follows: (singledual) / single. A linear regression model was fitted where the outcome was dual-task cost and predictors were age, sex, education and mutation status (carrier vs. non-carrier). As before, the model was refitted in mutation carriers alone with an additional predictor of years to expected onset.
Dual-task cost for mutation carriers and non-carriers was as follows (mean (SD)): carriers = 0.14 (0.24); non-carriers = 0.15 (0.20). This means that tracing speed in the dual task was ~15% slower than in the single task. There was no group difference between mutation carriers and non-carriers (regression coefficient 0.03 [95% CIs -0.16, 0.21], p = 0.78). Among mutation carriers, there was a trend towards an association between closer proximity to expected onset and greater dual-task cost, but this was not statistically significant (regression coefficient = 0.028 per year [95% CIs -0.004, 0.061], p = 0.083). Table 1 in the main manuscript presents descriptive statistics for the four task outcomes in the FAD and Insight 46 cohorts. Wilcoxon rank-sum tests were used to test for statistically-significant differences between the two cohorts in the dual-task condition (as Insight 46 participants were not administered single-task circle-tracing).

Comparison of performance in the FAD and Insight 46 samples
The only outcome with a statistically significant difference was circle-tracing speed (number of rotations), where FAD participants performed significantly faster (z = 4.2, p < 0.0001). This difference is likely to be age-related (FAD participants were 31 years younger than Insight 46 participants on average), consistent with a previous circle-tracing study that compared younger and older adults (Vaportzis et al., 2014), but could also partly be due to the fact that FAD participants may have been advantaged by practice effects, as they completed six single-task circle-tracing trials prior to this.

Performance of FAD participants on standard neuropsychological tests
Wilcoxon rank-sum tests were used to compare the performance of mutation carriers and non-carriers on a comprehensive battery of standard neuropsychological tests covering memory, language, executive function, visuospatial function, and general cognitive ability. Overall, they performed similarly; the only measures where mutation carriers showed evidence of poorer performance were Graded Difficulty Arithmetic, Block Design, and Spatial Span (Supplementary Table 1). These are measures that have some commonalities with circle-tracing and serial subtraction in terms of the tasks demands: • Block Design requires participants to reproduce a pattern using blocks that have various colour patterns on different sides; as such, it requires spatial visualization and visuomotor coordination, somewhat similar to the spatial transformation demands of circle-tracing with indirect visual feedback.
• Spatial span is measured with a block-tapping task that requires participants to observe the researcher tapping a sequence of blocks (using an array of 9 identical spatially-separated blocks) and then to repeat the sequence either forwards (in the same order) or backwards (in reverse order). This is primarily a measure of visuospatial working memory, but the backwards condition is somewhat analogous to circle-tracing with indirect visual feedback in its requirement for spatial transformation.
• Graded Difficulty Arithmetic (addition and subtraction) has a 10-second cut-off for each response, so the outcome (number of correct responses) includes an element of speed as well as accuracy.
The effect sizes of the group differences on these tests may be compared by expressing the difference between the means of mutation carriers and non-carriers as a proportion of the standard deviation for the whole sample.
These effects sizes are all similar in magnitude: Graded Difficulty Arithmetic 0.90 SD; Spatial Span forwards 0.87 SD; Spatial Span backwards 0.82 SD; Block Design 0.90 SD. Comparing this with the difference in serial subtraction and circle-tracing accuracy reported in the main manuscript, this is similar to the effect size for serial subtraction rate (0.87 SD), but the effect size for circle-tracing accuracy is somewhat higher (1.09 SD). This suggests that the circle-tracing task is at least as sensitive as these other tasks to the cognitive consequences of preclinical Alzheimer's disease pathology in FAD mutation carriers.