Patterns of default mode network co‐activation in young APOE ε4‐carriers: An HCP replication study

Abstract

AbstractBackgroundFilippini et al. reported an increased default mode network (DMN) co‐activation during rest in young APOE ε4‐carriers [1]. This work has been of substantial importance, often used to illustrate the trajectory of DMN alterations in APOE ε4‐carriers from young adults and normal aging to mild cognitive impairment and AD. However, this study only assessed a small sample. Here, we aimed to replicate their finding in a substantially larger sample from the Human Connectome Project (HCP).MethodWe included participants from the HCP S1200 dataset. We followed the methods of [1] for structural MRI and resting‐state fMRI data, with the following deviations: instead of using voxel‐based morphometry to assess gray matter density, we assessed gray matter volume with FreeSurfer. To test for differences in DMN co‐activation and gray matter volume between APOE ε4‐carriers and non‐carriers, we used Permutation Analysis of Linear Models (PALM). PALM controls for biases induced by the family structure of the HCP sample. Results were family‐wise error rate corrected at p < 0.05.ResultWe included 167 APOE ε4‐carriers and 419 non‐carriers for the DMN analysis. In contrast to [1], we did not find significant differences in DMN co‐activation between carriers and non‐carriers. For the morphometric analyses, 243 carriers and 544 non‐carriers were included. No differences in brain volume were found between carriers and non‐carriers. For both analyses, including covariates (different combinations of age, sex, education level, head motion) did not alter the results.ConclusionWe could not replicate the early findings of [1] in a large sample. Our results suggest that solely APOE ε4‐carriership does not influence brain function at a young age. Future work will focus on the use of polygenic risk scores to study whether any genetic predisposition for AD influences brain function in young adults. Still, genetic predisposition for AD might not have any measurable effect decades before clinical expression of AD. Therefore, repeating the analysis in a healthy, middle‐aged cohort could inform us whether the influence of genetic predisposition for AD increases over time. Altogether, our replication study is an important step forward in unravelling AD development. Resource: (1) Filippini et al., PNAS, 2009.