Longitudinal versus Cross-sectional effects of Age on MEG Power Spectra Parameters: Implications for Normative Models and Brain Ageing

Abstract

Ageing produces changes in brain function. To map this process, researchers increasingly utilise Magnetoencephalography (MEG) to identify spectral shifts across large, age-diverse cohorts. However most such research has used cross-sectional designs, which may be confounded by generational differences and high inter-individual variability. To address these limitations, we analysed longitudinal resting-state MEG data from the Cam-CAN cohort, an adult lifespan cohort with a repeat MEG recording after approximately a decade. This design allowed us to contrast cross-sectional age differences between people (at baseline) with longitudinal age changes within people. Our analyses revealed four key findings. First, we observed consistency between baseline and longitudinal effects of increasing age on the slowing of the Alpha peak frequency. Second, we identified non-linear changes in Beta power, whereby power increased from youth to mid-life, but decreased from mid-life to late-life. These findings validate the use of normative charts for mapping these spectral components. Third, we detected some longitudinal effects, such as power reductions in Alpha and slowing of Beta peak frequency, that were not apparent cross-sectionally, underscoring the importance of intra-individual designs. Fourth, we did not find the commonly reported age-related flattening of the aperiodic exponent of the power spectrum, except in direct measures of cardiac activity. Together, these results demonstrate that while cross-sectional data capture major trends, longitudinal data are essential for isolating the true neurophysiological signatures of the ageing process.