Adult lifespan trajectories of neuromagnetic signals and interrelations with cortical thickness

Abstract

Oscillatory power and phase synchronization map neuronal dynamics and are commonly studied to describe the healthy or diseased brain. Yet, little is known about the course of these features from early adulthood into old age. Leveraging magnetoencephalography (MEG) resting-state data in a cross-sectional adult sample (n = 350), we probed lifespan differences (18-88 years) in connectivity and power and interaction effects with sex. Building upon recent attempts to link structure and function in the brain, we tested the spatial correspondence of age effects on cortical thickness and those on functional networks and a direct relationship at the level of the study sample. We found MEG frequency-specific patterns for lower- and higher-order brain regions and divergence between sexes in low frequencies. Connectivity and power exhibited distinct linear trajectories with age or turning points at midlife. Variability in cortical thickness with age corresponded to those in neuromagnetic activity, particularly in delta and beta oscillations, and showed coupling in either unimodal or transmodal regions depending on the frequency. Altogether, we provide a comprehensive overview of the topographic functional profile of adulthood that can form a basis for neurocognitive and clinical investigations. This study further sheds new light on how the brain's structural architecture relates to fast oscillatory activity in large-scale networks.

Extensive MEG time-series phenotyping unveils neural markers predictive of age

“…Consistently, brain state dynamics differed between visual areas and higher-order networks 58 across the adult lifespan and were linked to cognitive performance, with a reduction in older people being more strongly associated with shifts between these states than in young people. As previously mentioned, age-related changes in the visual cortex have also been reflected in phase-based connectivity across different frequency bands 4 , but the predictive power of phase-related measures was rather poor in our study and others 26 . This may be because estimations of the phase of M/EEG signals are generally susceptible to noise, making it a less robust metric than autocorrelation, which is well-normalized between -1 and 1.…”

“…Furthermore, we observed a decrease in autocorrelation in the temporal cortex, particularly in the age groups after 40 years of age. Again, previous studies have identified the temporal cortex as an area associated with strong M/EEG power increases across the adult lifespan 3,4 , with an increase over the adult lifespan.…”

“…We used cross-sectional open-access data from the Cambridge Centre for Aging and Neuroscience (Cam-CAN) 60 . From 650 resting-state MEG data sets and T1-weighted MR images available from phase two of the study 61 , we further analyzed a subset of cleaned data ( n = 350) as reported elsewhere in detail 4 . Our sample included 50 healthy individuals of each age decade between 18 to 88 years and was balanced by sex.…”

“…We then resampled the data to 300 Hz using Fieldtrip 38 and, after filtering (first order Butterworth high-pass filter at 1 Hz), segmented the data into segments of 10 s length. Further processing steps are described in detail here 4 . In short, we followed an automatic approach to identify and reject segments containing artifacts from muscle movements (see https://www.fieldtriptoolbox.org/tutorial/automatic_artifact_rejection/), low-pass filtered the data at 70 Hz (first order Butterworth) and used independent component analysis (ICA) to identify ocular and cardiac artifacts.…”

“…From childhood through middle age, age-related changes have also been observed in the gamma range 2 . From early adulthood to old age, connectivity primarily changes in visual areas 9 , partially irrespective of the frequency range 4 . Depending on the age range investigated, power or connectivity measures showed non-linear age trajectories 2 .…”