Changes in Cortisol but Not in Brain-Derived Neurotrophic Factor Modulate the Association Between Sleep Disturbances and Major Depression

Abstract

Sleep disturbance is a symptom consistently found in major depression and is associated with a longer course of illness, reduced response to treatment, increased risk of relapse and recurrence. Chronic insomnia has been associated with changes in cortisol and serum brain-derived neurotrophic factor (BDNF) levels, which in turn are also changed in major depression. Here, we evaluated the relationship between sleep quality, salivary cortisol awakening response (CAR), and serum BDNF levels in patients with sleep disturbance and treatment-resistant major depression (n = 18), and in a control group of healthy subjects with good (n = 21) and poor (n = 18) sleep quality. We observed that the patients had the lowest CAR and sleep duration of all three groups and a higher latency to sleep than the healthy volunteers with a good sleep profile. Besides, low CAR was correlated with more severe depressive symptoms and worse sleep quality. There was no difference in serum BDNF levels between groups with distinct sleep quality. Taken together, our results showed a relationship between changes in CAR and in sleep quality in patients with treatment-resistant depression, which were correlated with the severity of disease, suggesting that cortisol could be a physiological link between sleep disturbance and major depression.

Keywords: brain-derived neurotrophic factor; pittsburgh sleep quality index; salivary cortisol awakening response; sleep disturbance; treatment-resistant depression.

Figure 1

Mean ± standard error of Pittsburgh Sleep Quality Index (PSQI) in the sleep disturbance (SD: n = 18 depressive patients), poor sleep (PS: n = 18 healthy volunteers) and good sleep (GS: n = 21 healthy volunteers) groups. ***p < 0.001, general linear model (GLM) and Fisher post hoc. The number 4 on first bar indicates volunteers that have score higher than 20.

Figure 2

(A) Box plot (median ± Q75% and Q25%) of last month sleep latency (in hours) according to the Pittsburgh Sleep Quality Index (PSQI) in the sleep disturbance (SD: n = 16), poor sleep (PS: n = 18) and good sleep (GS: n = 21) groups. ***p < 0.001, Kruskal–Wallis test. (B) Mean ± standard error of the previous month sleep duration (in hours) according to the PSQI, in SD (n = 16) PS (n = 18), and GS (n = 21) groups. ***p < 0.001 and **p < 0.05 one-way analysis of variance (ANOVA) and Fisher post hoc. Two outlier patients of SD were excluded from both statistical analyses.

Figure 3

Box plot of daytime sleepiness (median ± Q75% and Q25%), according to the Epworth Sleepiness Scale (ESS), in the sleep disturbance (SD: n = 18), poor sleep (PS: n = 18) and good sleep (GS: n = 21) groups. p ≥ 0.05, Kruskal–Wallis test.

Figure 4

Mean ± standard error of sleep duration (in hours) at D1, in the sleep disturbance (SD: n = 18), poor sleep (PS: n = 18) and good sleep (GS: n = 21) groups. p ≥ 0.05, one-way ANOVA.

Figure 5

Mean ± standard error of salivary cortisol awakening response (CAR: cm³) in the sleep disturbance (SD: n = 18), poor sleep (PS: n = 18) and good sleep (GS: n = 21) groups. ***p < 0.001, general linear model (GLM) and Fisher post hoc. Two patients in SD showed close values, one of them being superimposed on the graph.

Figure 6

(A) Mean ± standard error of serum blood brain-derived neurotrophic factor (BDNF) levels (pg/ml) in the sleep disturbance (SD: n = 18), poor sleep (PS: n = 18) and good sleep (GS: n = 21). p ≥ 0.05, ANCOVA. (B) Statistically significant Spearman correlation between serum BDNF levels and platelets (rho = 0.29, p ≤ 0.05).