Brain age prediction in subjects with sleep deprivation based on the ENIGMA-Sleep data

Sleep is essential for the maintenance of human physical health (Grandner et al., 2012; Reid et al., 2006), cognitive performance (Ferrie et al., 2011; Leng et al., 2017), and mental health (Freeman et al., 2017; Joao et al., 2018). Experimental sleep deprivation and different degrees of sleep restriction permit the characterization of human sleep behaviors and brain's responses to inadequate sleep (Durmer and Dinges, 2005; Elmenhorst et al., 2017; Elmenhorst et al., 2018; Van Dongen et al., 2003). Sleep disruption is known to impact not only brain functions but also affects brain anatomy; earlier studies found that sleep deprivation is associated with lower volumes of gray matter across various brain regions (Åkerstedt et al., 2020; Liu et al., 2014; Long et al., 2020; Sun et al., 2020), however with inconsistent localization of effects. Other studies reported alterations in white matter microstructure (Elvsåshagen et al., 2015; Voldsbekk et al., 2021), increased ventricles size, and cognitive decline (Lo et al., 2014). In addition, one-night sleep deprivation was found to increase the Aβ burden in the right hippocampus and thalamus of healthy controls (Shokri-Kojori 2018), suggesting an interrupted clearance pathway due to sleep deprivation. These prior studies suggest that sleep deprivation may impact anatomical features of the brain, but there is a need for large-scale replications and extensions of previous work. Increased brain age compared to chronological age is a risk factor for several neurodegenerative and psychiatric disorders (Wyss-Coray, 2016). Brain-age prediction is a widely studied topic that aims to estimate the trajectory of brain aging (Franke et al., 2019), and its association with pathological conditions. MRI-derived features used to predict individuals’ brain age have proven to constitute a proxy for overall health (Habes et al., 2016, Koutsouleris et al., 2014, Cole et al., 2018). Sleep patterns also change with aging. In particular, older adults tend to sleep less, and present a more fragmented sleep with lower slow-wave sleep. They also present a reduced sleep rebound following sleep deprivation, report less sleepiness under acute sleep deprivation conditions, and report a smaller increase in lapses of attention after sleep deprivation [23]. However, little is known on whether brain-age models reflect the impact of sleep deprivation on the brain structure - i.e., the difference between predicted age and real age (so-called brain-age delta) using MRI data. We hypothesized that the morphological alterations associated with sleep deprivation would lead to an increased brain-delta score.