Authors:
Yu-Shiuan Lin (Basel | CH)
Janine Weibel (Basel | CH)
Hans-Peter Landolt (Zurich | CH)
Francesco Santini (Basel | CH)
Corrado Garbazza (Basel | CH)
Martin Mayer (Basel | CH)
Helen Slawik (Basel | CH)
Stefan Borgwardt (Basel | CH)
Christian Cajochen (Basel | CH)
Carolin Reichert (Basel | CH)
Aim: Caffeine is the most commonly used psychostimulant worldwide. Underlying its well-known benefit on vigilance, caffeine dampens the accumulation of sleep homeostatic pressure. Increased sleep pressure, e.g. prolonged wakefulness or fragmented sleep, is associated with lower cerebral grey matter (GM) volume1-3 and cell death4-6. We are therefore interested in whether daily caffeine intake leads to changes in GM through its regular influence on sleep homeostasis. Methods: In our double-blind crossover study, each of the 20 male caffeine consumers (26.2± 4.1 y/o, daily intake 471.5± 112.24 mg) completed caffeine (3x 150 mg/day) and placebo for 10 days. Sleep polysomnography was performed the evening of day 9. GM was measured via T1-weighted MPRAGE (1x1x1mm3, TR=2000ms, TE=3.37ms) in a 3T fMRI scanner. Caffeine effect on voxel-wise GM was estimated by flexible factorial model. Due to the bias caused by caffeine-induced reduction in cerebral blood flow (CBF)7, 2D-EPI ASL (4x4x4mm3, TR=3000ms, TE=12ms) was performed, and voxel-to-voxel multimodal mixed model was applied to adjust coefficients. Sleep EEG slow wave activity (SWA, 0.75- 4.5hz) was quantified as the index of sleep pressure. AUC of caffeine and paraxanthine (AUC-CAPX) was calculated to quantify the substance exposure. To examine the verbal working memory related to the GM alterations, we examined the accuracy and reaction time (RT) in N-Back task. Results: In the caffeine condition, total GM was lower compared to placebo (p< .001). Voxel-wise analyses indicated that the most prominent GM decrease was in medial temporal lobe (mTL) including hippocampus (pFWE< .05). Lower total and mTL GM were associated with AUC-CAPX (p< .001). SWA did not differ between conditions and exerted no mediation on caffeine-associated GM reduction. However, the variance of total GM was fully accounted by the total CBF reduction, but not mTL CBF to mTL GM. N-back data indicated a worse accuracy and longer RT during rejection response (missed and correct rejection, pall< .05) in caffeine condition, while AUC-CAPX was positively associated with accuracy and RT (pall< .05). Conclusion: While total GM estimation is confounded by the variance of total CBF, hippocampal GM is reduced during daily caffeine consumption independent of SWA and hippocampal CBF. Verbal working memory is better at higher acute caffeine exposure but worse in daily intake, that implies the distinction between acute and long-term effects of caffeine.