This open-label trial was conducted at the Federal University of Rio Grande do Norte (UFRN) between 2018–2019. It has been approved by the UFRN Human Ethics Committee (# 2628,202) and registered at http://clinicaltrials.gov (U1111-1215-4472). The procedures of this study comply with the ethical standards of the relevant national and institutional committees for human experimentation and with the Declaration of Helsinki of 1975, revised in 2008. All subjects provided written informed consent prior to participation and it was ensured freedom to withdraw from the study at any time and without any prejudice. All study’s information has been kept confidential.
Volunteers
The recruitment of volunteers was performed by advertising on radio, social and academic media. All subjects have undergone a full clinical evaluation performed by a trained psychiatrist including anamnesis and mental health evaluation assessed by Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) and the Hamilton Depression Assessment Scale. After screening, the volunteers were grouped as follows:
Patients group (PG): 30 volunteers (16 women and 14 men) with mild to moderate depressive episode without psychiatric, neurological, or physical (metabolic and inflammatory) comorbidities were included in this group and selected to start the treatment. Other exclusion criteria were the previous use of antidepressant and current use of antidepressants, anxiolytic, corticoids, or drugs with action on neurovegetative functions, mood, and cognition. Therefore, the patients were drug-free during this study.
Control group (CG): 25 healthy volunteers (13 women and 12 men) with similar socio-demographic characteristics of the patients group, without history or current diagnosis of physical (metabolic and inflammatory), neurological or mental disease, and no use of regular medication with action on neurovegetative functions, mood and cognition. This group did not receive treatment or intervention.
Study design
All volunteers (PG and CG) slept, individually, one night at the Laboratory of Neurobiology and biological rhythms of UFRN, when the Pittsburgh Sleep Quality Index (PSQI) [42] was assessed. In the following morning, around 6 a.m., the saliva and blood were collected to measure the salivary cortisol awakening response (CAR) and serum cortisol, respectively. Prior the patients started the treatment, the psychometric scales were assessed: Beck Depression Inventory (BDI), Beck Anxiety Inventory (BAI) and Rosenberg Self-Esteem Scale (RSE). Over the following 16 weeks, patients underwent the treatment. After that, up to one week from the end of treatment, patients slept once again in the sleep lab. Sleep quality was assessed by PSQI and blood and saliva were collected in the following morning. A new clinical evaluation with the same psychiatrist was conducted and the psychometric scales (BDI, BAI and RSE) and HAM-D were assessed.
Treatment
Patients were treated with group Cognitive-Behavioral Therapy (gCBT) in monotherapy for 16 weeks; 12 weekly sessions and two fortnightly sessions (reinforcement and closing sessions). The duration of each session was about 2 hours and patients comprised 3 therapeutic groups with 10 individuals per group, in average. The therapy protocol was adapted from [43] and [44], and applied by a team headed by a trained psychologist with expertise in the cognitive-behavioral approach, a co-therapist, and an observer.
As an ethical consideration, individual sessions were held in case of decompensating or risk of suicide. On average, there was one individual session per patient. At the end of the study patients that had not achieved remission were included as outpatients in Psychology Service or/and psychiatry at Hospital Universitário Onofre Lopes, both from UFRN.
Instruments
Hamilton Depression Assessment Scale [45] was used to MDD diagnosis, to assess its severity and its remission. It is a semi-structured interview for identification of frequency and intensity of depressive symptoms performed by a trained psychiatrist. HAM-D is one of the most usual tools to assess depressive symptoms and, because of that, was adopted as the primary outcome measure in this trial [46, 47]. HAM-D has 4 categories of classification: a) mild major depression 10 < scores < 13; b) moderate: 14 < scores < 17 and c) severe: scores > 17 [48]. The Beck Depression Inventory [49] is a 21 items, self-reported instrument that assesses depressive symptoms during the last week and proposes their classification in four levels: “Minimal” (0–11 scores), “mild” (12–19 scores), “moderate” (20–35 scores) and “severe” symptoms (36–63 scores). It was validated for adult Brazilian clinical population [50] (α = 0.93). The Beck Anxiety Inventory [51] is a 21 questions, self-completion scale, which measures somatically, affectively and cognitively the anxiety level during the last week and proposes its classification into 4 levels as following: “minimum” (0–10 scores), “mild” (11–19 scores), “moderate” (20–30 scores) and “severe” (31–63 scores). This instrument was validated to adult Brazilian population and was proved suitable for use in clinical population (α = 0.83–0.92) [52]. Rosenberg Self-Esteem Scale [21] is a 10 items self-completion instrument to measure self-concept traits. The total score ranges from 0 to 40. It was validated to adult Brazilian population (α = 0.90) [53]. In this study conventional or non-inverted correction was adopted. Pittsburgh Sleep Quality Index (PSQI) is a self-assessment instrument that has 7 components and it is used to measure sleep quality. The overall score ranges from 0 to 21 points, in which can be categorized into good sleep quality (0–4 points), poor sleep quality (5–10 points) and suggestive of sleep disorder (greater than 10 points). This instrument was validated to adult Brazilian population [54] (α = 0.82).
Collection and dosage of biological material
The salivary cortisol awakening response is a feedforward mechanism that prepares the individual to daily activities, therefore has minor influence of acute stressors. CAR changes have been associated with dysfunction of the HPA axis, observed in some mental illnesses [55, 56]. On the other hand, Serum cortisol is highly influenced by an enormous sort of stressors and suffers from time-of-day modulation, thus its measurement has an acute value. Therefore, these two measures were select and analyzed aiming at helping in the consolidation of cortisol as biomarker of MDD [38, 57].
To account for the circadian oscillation in cortisol levels, both saliva and blood were always collected about 6 a.m. Volunteers were in fasting for approximately 8 hours. Using Salivette® devices (Sarstedt Numbrecht, Germany), two saliva samples were collected by the volunteers under the researcher supervision. The 1st collection was done immediately at awakening and the 2nd was done after 30 minutes. During collection participants were restricted to bed and were instructed to rest and not eat or drink. Blood collections were performed, using disposable perforating material (needle and syringe), immediately after saliva collections by trained laboratory technicians or researcher. Salivary cortisol was measured by DRG® Salivary Cortisol ELISA kit SLV-4635 and Serum cortisol by DRG® Cortisol ELISA Kit 1887.
Statistical analysis
Depressive (BDI) and anxiety symptoms (BAI), sleep quality (PSQI), self-esteem (RSE), serum cortisol (SC) and cortisol awakening response (CAR) were the quantitative dependent variables evaluated in this study. Groups (PG and CG) and patient’s remission (remitted: R and non-remitted: NR) were considered categorical independent variables. Sociodemographic characteristics, such as gender, age, education, and income were investigated as covariates. The CAR was calculated from the change (%) in salivary cortisol between 0 and 30 minutes after awakening [56].
The effect of sociodemographic characteristics was assessed by Wilcoxon sum rank test (for age) and chi-square test (for gender, education level and income). The effect of intervention was assessed by Wilcoxon sign rank test both considering the patients group (PG) as a whole, as well as stratifying it by remission condition (remitted patients [R] and non-remitted [NR]). In order to check whether the changes achieved with the intervention were comparable with a healthy pattern, we performed a Wilcoxon sum rank test comparing the baseline and post-treatment values to a healthy control. The effect sizes (r) and its bootstrapped confidence interval (CI 95%, 1,000 resamples) are reported as r [lower limit, upper limit], and were obtained from the rcompanion package. The correlations between biomarkers after the treatment were assessed for remitted and non-remitted patients using Spearman’s correlation (⍴). Finally, a binary logistic regression was performed to find potential baseline predictors of remission, using non-remitted patients as reference (NR = 0; R = 1). Nagelkerke’s pseudo-R2 and OR is reported as importance measures of the prediction.
All analyzes were performed using R (4.0.2), assuming a significance level of p < .05.