Can an online battery match in‐person cognitive testing in predicting age‐related cortical changes

Background

Studies into cognitive changes typically employ detailed in person cognitive testing, which is not always feasible. Therefore, we compared the relationship between brain morphology (sulcal width) and cognitive functioning, using an online and an in‐person modality and disentangled the influence of age, sex, β‐Amyloid (Aβ) and APOE‐status.

Method

141 healthy participants (mean age 60, range 46‐71 years, 75% female) assessed with structural MRI; cognitive batteries both, face‐to‐face and online (Cambridge Brain Systems), Aβ status (Fluorine‐18 florbetaben‐PET scan) and APOE genotype (Lupton et al., 2021); Canonical Partial Least Square method to compare cognitive modalities and Sulcal width (SW; Morphologist pipeline‐BrainVISA toolbox; Borne et al., 2020). Age effects tested with two‐sided Wald Test. Analysis of covariance to test age and sex‐interactions, amyloid and APOE status, sex‐effects (controlling for age), Aβ, and APOE (controlling for age and sex).

Result

The single robust mode for brain (SW) ‐ behaviour (cognition) covariation loaded most strongly onto memory and executive functions for both the onsite (1st mode, p = 0.013, cov = 3.55, z‐cov = 2.93, R² = 0.18, z‐R² = 0.95; 2nd mode, p>0.99), and the online battery (1st mode, p<0.001, cov = 2.76, z‐cov = 4.71, R² = 0.14, z‐R² = 1.15; 2nd mode, p = 0.99) (Fig. 1).

Cognition‐related SW showed a regional pattern similar for online and in person cognitive appraisal. Significant effect of sex on SW projections in both the online and onsite conditions with larger sulcal width for men (p<0.001) and a similar effect on onsite cognitive projections with lower performance for men (p<0.001) (Fig. 2a). Cognitive performance ‐ both online (p = 0.03) and in‐person (p<0.01) – showed a significantly steeper decline with age for Aβ‐positive participants (Fig. 2b). Variance explained for the online cognitive assay (R² = 0.15) was only slightly less than for the in‐person testing (R² = 0.18). Brain‐behaviour z‐transformed covariance was likewise comparable across modalities.

Conclusion

Similar sulcal width brain projections for both cognitive modalities, with memory and executive domains showing the strongest loadings. Aβ‐aggregation associated with a steeper cognitive projection slope for both batteries, suggesting that in our preclinical sample the early stages of Aβ accumulation accelerate cognitive ageing potentially before translation into structural brain changes. Adequate sensitivity of online cognitive tests for studying age‐related neurobiology of cognition is suggested.