DOI: 10.1002/alz.082212 ISSN: 1552-5260

Age differentially mediates the relationship between plasma biomarkers and cognition detected by digital screening

Sofia Toniolo, Sijia Zhao, Anna Scholcz, Benazir Amein, Anne Gallacher, Akke Ganse‐Dumrath, Bernard R Bukala, Sian Thompson, Sanjay Manohar, Henrik Zetterberg, Masud Husain
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology



Plasma biomarkers are emerging as a useful tool to measure neurodegeneration, particularly in patients with Alzheimer’s Disease (AD). Digital cognitive testing is also increasingly used to detect cognitive changes in AD. Age is a major determinant of performance at cognitive testing and can affect the levels of different plasma biomarkers. However, little is known about how age mediates the relationship between plasma biomarkers and cognition.


38 AD and 48 age‐matched controls were recruited from the Oxford Centre for Cognitive Disorders. We used a fully remote digital visual short‐term memory (STM) task (Figure 1) and extracted eight cognitive metrics (Figure 1): Identification Accuracy (percentage of correctly identified items), Absolute Error (distance from the response location to the original item), Target detection (probability of correctly identifying the target), Misbinding (erroneously localizing an item to the remembered location of another item in memory), Guessing (random guessing), Identification and Localization Time (time to identify and drag the correct object to its remembered location), and memory Imprecision (width of the distribution around the target). Plasma p‐tau 181, GFAP, NFL, Aß40 and Aß42 were measured, and the Aß42/40 ratio was calculated. In Model 1 (Figure 2) the correlation between digital cognitive metrics and plasma biomarkers is shown, while in Model 2 age was regressed out. A mediation analysis was then performed to look at the effect of age as a mediator.


In Model 1, most cognitive metrics on the STM task were strongly associated with p‐tau 181, GFAP and NFL. Identification Accuracy was also negatively correlated with the Aß42/40 ratio. No association with Aß42 or Aß40 was found. However, regressing out age had a profound impact on these correlations, particularly for GFAP and NFL, while p‐tau 181 was not significantly affected (Figure 2 and 3). The mediation analysis (Figure 3), shown for Identification Accuracy, confirmed that age had a more profound impact on GFAP and NFL compared to p‐tau 181.


These new digital metrics show a strong correlation with plasma biomarkers. Finally, age is an important mediator between plasma biomarkers and cognition, but that varies across different biomarkers.

More from our Archive