The remote Mobile Toolbox for capturing cognitive change in preclinical Alzheimer’s disease

BACKGROUND: Remote, smartphone-based cognitive assessments such as the Mobile Toolbox (MTB) may facilitate the accessibility and scalability of Alzheimer's disease (AD) research and clinical trials. We previously showed that a composite of MTB memory and processing speed tests provides a feasible approach to measure cognitive processes that are relevant in preclinical AD. Here, we examined whether this MTB-composite is sensitive to amyloid- and tau-related cognitive change over time. 
METHOD: The MTB was deployed longitudinally in N=100 cognitively unimpaired (CU) older adults from four affiliated observational cohort-studies. All participants self-completed the MTB remotely on their personal mobile device. Our MTB-composite of interest includes two measures of episodic memory (Arranging Pictures and Faces & Names) and two measures of attention and processing speed (Sequences and Number-Symbol Match). Amyloid-β positron emission tomography (PET) and tau PET were available within -1.56±1.97 years of the first MTB assessment. Global amyloid-β burden was dichotomized into negative (Aβ-) versus positive (Aβ+) using a distribution volume ratio cut-off 1.19. Tau deposition was measured as standardized uptake value ratios (SUVR) in medial-temporal lobe (MTL) and neocortical (NEO) regions. Linear mixed effect models correcting for age, sex, and years of education were used to investigate whether Aβ-status (dichotomous) and MTL and NEO tau deposition (continuous, normalized SUVR values) were associated with change over time on the MTB-composite. 
RESULT: N=100 CU participants (age 70.5±7.8, 63% female, 16.8±2.5 years of education, 27% Aβ+) completed MTB baseline assessments. Of those, n=91 had at least one follow-up assessment available (7.9±2.4 months after baseline) and n=64 had a second follow-up assessment (14.6±3 months after baseline). Compared to Aβ- individuals, the Aβ+ group showed greater decline in MTB-composite scores (Time*Aβ+ status: β = -0.43, 95%CI[-0.72 - -0.14], p=.005, Figure 1). Greater MTL and NEO tau deposition were also associated with greater decline in MTB-composite scores (Time*MTL tau: β = -0.30, 95%CI[-0.48 - -0.13], p=.001, Time*NEO tau: β = -0.20, 95%CI[-0.38, -0.01], p=.042, Figure 2). 
CONCLUSION: These initial findings suggest that the MTB can capture AD-related cognitive decline and thereby may provide a promising tool to remotely monitor cognition in preclinical AD research and clinical trials.