Effect of sleep apnea and insomnia on the association of depression with quantitative electroencephalogram measures (QEEG) in adult men during sleep – the MAILES study

Introduction: Quantitative EEG (qEEG) abnormalities are present in 80% of patients with psychiatric disorders. Small studies of resting, awake qEEG in patients with depression show variation in findings. Both increase and decrease in slow wave activity has been reported in depressed patients in addition to increased alpha and beta activity. We have previously shown co-morbid sleep apnea and insomnia have additive effects on depression prevalence and severity. We aimed to determine the effect of sleep apnea and insomnia symptoms on the relation between sleep qEEG parameters and depression in a large sample of community dwelling men. Materials and methods: Data were drawn from a randomly-selected cohort of men aged ≥40y at recruitment (2002-5) in Adelaide, South Australia. (MAILES study). Of the 837 men with no prior OSA diagnosis who underwent full in-home polysomnography (Embletta X100; 2010-11), the C3 EEG tracings in n=635 men with known depression status were subjected to quantitative power spectral analysis. PSG variables were log transformed to normalise the distribution for regression against AHI adjusted for age. EEG short and longwave densities consisted of alpha/sigma/beta and delta/theta waves respectively, and distinguished between stages (S2 v SWS v REM). The axes variation across the cohort determined by principal component analysis (PCA). Depression symptom scores were regressed onto log transformed short and longwave EEG measures adjusting for age, AHI, BMI, financial stress, insomnia symptoms, marital status and medication. A sensitivity analysis was performed using linear regression of depression scale scores to confirm findings. Results: EEG measures were moderately correlated (median=0.5, r=0.2-0.8). PCA indicated 3 major axes of variation: (PC1) activity level, (PC2) REM vs SWS and (PC3) long v short. Linear regressions indicated associations with PC1 (p=0.003) and PC3 (p=0.05) indicating depression more prevalent in individuals with elevated shortwave densities. Repeating the analyses with short and longwave measures separately confirmed this conclusion. Short wave densities, irrespective of stage, were associated with depression while long wave were not. Other factors associated with depression in this analysis were elevated AHI (p=0.01), financial stress (p=0.002), having a partner (p=0.04), medication (p< 0.001) and insomnia (p<0.001), but not age (p=0.21) nor BMI (p=0.78). EEG measures explain a greater proportion of the variance in outcome compared to AHI (full model R2=23.8%, excluding AHI R2=23.0%, excluding EEG R2=22.1%). Conclusions: Depression was associated with increased fast wave activity in community dwelling men. After further adjustment for insomnia symptoms and AHI, long-wave (slow) activity was no longer associated with depression. qEEG from sleep studies may add useful information to AHI in identifying sleep apnea clinical phenotypes at-risk for depression. Power spectral analysis data are recorded in sleep studies although not routinely reported, and can be easily accessed using available algorithms.