BASELINE SLEEP PREDICTS ANXIETY PHENOTYPES AND POST-ANAESTHETIC COGNITIVE PERFORMANCE DIFFERENTLY IN ANXIETY-CLASSIFIED AND STRESS-REACTIVE MICE

Preoperative anxiety (POA) correlates with postoperative delirium (POD), yet their neurophysiological relationship remains unclear. We investigated whether baseline sleep/EEG features predict anxiety phenotypes and modulate post-sevoflurane cognitive and sleep outcomes, comparing behaviourally anxiety-classified wildtype mice with stress reactivity (SR) mice, selectively bred for extremes in hypothalamic-pituitary-adrenal (HPA) axis reactivity. Cohort 1 (n=50 male C57BL/6N) received EEG/EMG implants and 24-hour baseline recordings before cued fear conditioning (FC) classified mice into high anxiety (HA) or low anxiety (LA) groups via k-medoids clustering. Cohort 2 (n=25) underwent identical anxiety phenotyping, followed by Water Cross Maze (WCM) training. After an experimental sevoflurane anaesthesia (1h, 2.0%vol, EEG-guided), WCM re-testing, and fine-grained behavioural syllable analysis using DeepLabCut and Keypoint-MoSeq was performed. SR mice (high/intermediate/low stress reactivity mice [HR/IR/LR]; n=12 males each) underwent identical WCM/anaesthesia/sleep monitoring procedures without FC. Baseline sleep analysis revealed contrasting phenotype-sleep relationships: HA mice exhibited diminished rapid eye movement sleep (REMS) and elevated NREMS, whereas HR showed elevated REMS, compared to LR post-anaesthesia, HA mice displayed subtle cognitive alterations (prolonged confusion syllables, decreased velocity, attenuated motor responses). In contrast, all three SR lines maintained robust WCM performance post-anaesthesia. However, LR mice exhibited reduced REMS, increased wakefulness, prolonged REMS latency accentuating baseline differences, while HR mice showed preserved sleep architecture. Behavioural anxiety and HPA-axis reactivity phenotypes show opposing baseline sleep patterns and differentially modulate post-anaesthetic outcomes. High anxiety contributes to cognitive impairments, while low stress reactivity interferes with sleep recovery, which may reveal distinct neurobiological pathways underlying perioperative vulnerability.