BDNF and proBDNF Serum Protein Levels in Obstructive Sleep Apnea Patients and Their Involvement in Insomnia and Depression Symptoms

Introduction: Obstructive sleep apnea (OSA) is a disorder that, apart from somatic sequelae, increases the risk of developing psychiatric conditions. Brain-derived neurotrophic factor (BDNF) signaling pathway is involved in the pathophysiology of depression and insomnia. Therefore, the study aimed to investigate differences in concentrations of BDNF and proBDNF in patients with OSA and healthy individuals, to evaluate diurnal changes of these proteins, and to assess the correlations with psychiatric symptoms. Methods: Sixty individuals following polysomnography (PSG) were divided into two groups based on the apnea-hypopnea index (AHI): OSA patients (AHI ≥ 30; n = 30) and control group (AHI < 5; n = 30). Participants filled out questionnaires: Beck Depression Inventory (BDI), Athens Insomnia Scale (AIS), and Pittsburgh Sleep Quality Index (PSQI). Peripheral blood was collected before and after PSG. Protein concentrations were measured using ELISA. OSA group was divided into subgroups: AIS (−)/AIS (+) (AIS > 5), PSQI (−)/PSQI (+) (PSQI > 5), and BDI (−)/BDI (+) (BDI > 19). Results: No differences in BDNF and proBDNF protein levels were observed between OSA and the control groups. However, BDNF and proBDNF evening protein concentrations were higher in the AIS (+) and PSQI (+) groups (p < 0.001 for all). The BDI (+) group was characterized by lower morning levels of both proteins (p = 0.047 and p = 0.003, respectively). Conclusions: BDNF signaling pathway might be involved in the pathophysiology of depression and insomnia in patients with OSA. BDNF and proBDNF protein levels might be useful in defining OSA phenotypes.


Introduction
Obstructive sleep apnea (OSA) is a common sleep-related breathing disorder characterized by recurrent pauses in breathing during sleep due to partial or complete obstruction of the upper airways [1]. Numerous pauses in breathing conduce to intermittent hypoxia (IH), which is closely correlated with systemic inflammation and oxidative stress [2]. Those factors might contribute to OSA comorbidities, like hypertension, other cardiovascular diseases, or metabolic disorders, especially type II diabetes mellitus [3][4][5][6][7][8]. Studies have also highlighted the association between conditions with immune-mediated background and OSA, which indicates some immune system dysregulation occurring in the course of OSA in susceptible individuals [9][10][11][12][13].
Recently more studies have been focusing on the psychiatric aspects of this disorder, i.e., increased prevalence of affective diseases (particularly depression), cognitive impairment, or co-existing insomnia. The prevalence of this condition in patients with OSA in some studies reaches up to 84% [14]; it might greatly affect a patient's prognosis and overall health as well as affect adherence to treatment and its potential effectiveness [15][16][17]. In one study, the risk of all-cause mortality in the OSA group with comorbid insomnia was 47%

Polysomnography
Patients were admitted to the sleep lab at 21:00 h (±0.5 h) and underwent physical examination (measurement of body mass, height, heart rate, and blood pressure). Standard nocturnal polysomnography was performed by recording the following channels: electroencephalography (C4\A1, C3\A2), chin muscles and anterior tibialis electromyography, electrooculography, measurements of oronasal airflow (a thermistor gauge), snoring, body position, respiratory movements of chest and abdomen (piezoelectric gauges), unipolar electrocardiogram and hemoglobin oxygen saturation (SpO 2 ) (Alice 6, Phillips-Respironics). Sleep stages were scored according to the criteria based on the 30-s epoch standard [34]. Apnea was attained with over 90% the airflow reduction for at least 10 s. Hypopnea was defined as at least a 30% reduction of airflow for at least 10 s, accompanied by over a 3% decrease in SpO 2 or arousal. Encephalography arousals were scored according to the American Academy of Sleep Medicine guidelines [34].

Questionnaires
Questionnaires included three research instruments: Pittsburgh Sleep Quality Index (PSQI), Beck Depression Inventory (BDI), and Athens Insomnia Scale (AIS). They were filled in by each participant in the morning after polysomnography.

Pittsburgh Sleep Quality Index (PSQI)
Self-evaluation questionnaire assessing seven different aspects of sleep in adults. It evaluates sleep quality parameters such as difficulties with falling asleep, problems with maintaining continuity of sleep, functioning during the day, and questions regarding the most frequent causes of sleep disorders over the past four weeks. They all make up the outcome, assessed from 0 to 21 points. Results higher than 5 points indicate low sleep quality and differentiate patients into "poor" and "good" sleep [35][36][37]. This principle was used to determine the cutoff point used to divide patients with OSA into two groups: the PSQI (−) group (PSQI ≤ 5) and the PSQI (+) group (PSQI > 5). PSQI proved to have high internal consistency, as indicated by Cronbach's alpha of 0.83 [38]. A validated PSQI version in Polish was used in the study [39].

Athens Insomnia Scale (AIS)
A questionnaire consisting of 8 questions dedicated to insomnia studies. The first five questions are according to the ICD-10 criterion of insomnia diagnosis, including assessing difficulty with sleep induction, awakening, total sleep time, and overall quality of sleep. The last three items evaluate day consequences of insomnia, such as subsequent day wellbeing, functioning, and daytime sleepiness [46]. Each question is scored from 0 to 3 points, corresponding to "no problem at all" to "very serious problem", respectively. Summarizing all items assessed to 24 points in total. Saldatos et al. reported that the cutoff score is 5.5 points in European countries [46], and it was used in the study to determine the AIS (−) group (AIS ≤ 5) and AIS (+) group (AIS > 5). A validated AIS version in Polish was used in the study [47].

Blood Collection and Protein Level Assessment
Peripheral blood samples were collected in the evening before and in the morning following PSG examination into collection tubes with a clot activator. Blood samples were centrifuged immediately following the blood draws at 4 • C. The serum was collected and stored at −80 • C. The serum BDNF and proBDNF protein concentrations were assessed by ELISA kit (Human BDNF (Brain-Derived Neurotrophic Factor) ELISA Kit and Human pro-BDNF (pro-Brain-Derived Neurotrophic Factor) ELISA Kit respectively, FineTest, Wuhan, China). The absorbance was measured at λ = 450 nm wavelength by an absorbance reader (BioTek 800 TS, Agilent Technologies, Santa Clara, CA, USA).

Statistical Analysis
Statistical analysis was performed at a significance level of 0.05 using two-tailed tests. The normality of the distribution of variables was tested with the Shapiro-Wilk test. For variables with a normal distribution, the data is presented as the mean with the standard deviation; for variables with a distribution other than normal, the data is presented as the median with the interquartile range (IQR). Chi-square and Chi-square tests with Yate's correction were used to assess nominal variables in situations where the size of the smallest group was, respectively: above 15 and in the range of 5-15. Comparisons of independent groups were made using the student's t-test (for variables with a normal distribution) and the Mann-Whitney U test (for variables with a different distribution than normal). Dependent groups were compared with the t-student test for dependent variables (for variables with a normal distribution) or Wilcoxon (for variables with a different distribution than normal). Correlations between continuous variables were tested with Spearman's rank correlation test. The analysis was performed using IBM SPSS Statistics version 28 (2021, Armonk, NY, USA).

Results
Baseline characteristics and comparison between the control group (n = 30) and OSA group (n = 30), including demographic data, polysomnography parameters, protein concentrations, and questionnaire results, are shown in Table 1.
No differences were found between the morning and the evening protein concentrations in the case of BDNF (p = 0.162) and proBDNF (p = 0.791) in all participants of the study; similarly, no differences were observed in the control group (p = 0.232 and p = 0.439 respectively) and the OSA group (p = 0.624 and p = 0.821 respectively) ( Figure 1).
In the OSA group, strong positive correlations between the morning and the evening of BDNF (r = 0.580, p < 0.001) and proBDNF (r = 0.527, p = 0.003) concentration were observed. Moreover, a very strong positive correlation between BDNF and proBDNF protein concentration in the morning (r = 0.860, p < 0.001) and in the evening (r = 0.923, p < 0.001), respectively, was achieved. Additionally, BDNF protein concentration positively correlated with total sleep time both in the evening (r = 0.386, p = 0.035) and in the morning (r = 0.412, p = 0.024) ( Figure 2).

Discussion
BDNF signaling pathway might be a contributor to the course of OSA. It seems to influence the risk of developing OSA comorbidities, such as insomnia, cognitive impairment, and depression. In this study, we demonstrated increased evening levels of BDNF and proBDNF in patients with OSA who scored high on questionnaires assessing poor sleep quality and insomnia symptoms (PSQI and AIS, respectively). Moreover, the severity of depression symptoms assessed with BDI was associated with decreased morning serum levels of both proteins in the OSA group. This may suggest the plausible involvement of those proteins in developing mood and sleep disorders in OSA. In addition, serum levels of BDNF and proBDNF are not significantly different between healthy individuals and patients with severe OSA, and no diurnal changes are present.
The majority of evidence indicates that intermittent hypoxia, one of the most damaging effects of OSA, causes decreases in BDNF levels in animal models [48][49][50][51][52][53][54]. Fang et al. studied the influence of CIH (chronic intermittent hypoxia) on neurodegeneration of the optic nerve in mice model, showing decreased BDNF levels. Those changes were reversible after 7,8-dihydroxyflavone (7,8-DHF) administration, an antioxidant [48]. After 7,8-DHF application, oxidative stress was reduced, and BDNF/TrkB/CREB pathway increased in activity, which emphasized the role of reactive oxygen species (ROS) in the impairment of BDNF expression [21,55]. Mice and piglet model of OSA confirmed the involvement of Trk/CREB pathway regulation of the BDNF disruption [51,56]. Another study obtained similar results of BDNF pathway downregulation in the CIH mice model, but it emphasized the role of impaired iron metabolism as a potential pathomechanism [49].
Our study didn't find any significant differences in BDNF and proBDNF levels between OSA and the healthy control group. Interestingly, most human studies on BDNF in OSA showed no differences in BDNF levels [30][31][32][57][58][59]. On the other hand, Shah et al. showed increased BDNF expression in the soft palate muscles of snorers and patients with OSA [60]. A possible explanation of this paradox is a time of exposure to IH. In studies on animal models time of exposure was between 3 to 12 weeks. OSA is a chronic disease, and it has been affecting patients for many years. This time is sufficient to activate the adaptive mechanisms. It has been suggested that sensorimotor neuropathy may cause upper airway collapse in patients with OSA [61]. Shah et al. showed that neuromuscular injuries caused by vibrating are typical in those groups of patients [62]. The following denervation was correlated with increased BDNF expression in the local environment, the same with swallowing dysfunction. The authors concluded that it might be an adaptation to neuromuscular injuries, which can lead to reinnervation [60]. Flores et al. also observed increased BDNF levels in patients with OSA. They emphasized the role of BDNF in neuroprotection in patients with OSA. In their study, higher BDNF levels correlated with higher oxygen desaturation index and with the Montreal Cognitive Assessment questionnaire score; it suggests that an increase in BDNF concentration might result in an improvement in cognitive functions. Thus, BDNF could be considered as o protective factor against cognitive decline [33]. Arslan et al. also reported the protective role of BDNF in patients with OSA in response to neurodegeneration. In their study, they found increased levels in mild and moderate-to-sever OSA than in healthy participants [63]. Moreover, BDNF levels correlated positively with the hypoxia [63]. In the same study, a similar correlation was also received for hypoxia and neurofilament light chain level (NF-L) [63], which is used as a biomarker of axonal damage in the AD [64]. However, the correlation of BDNF with NF-L in each group was insignificant [63]. The impact of hypoxia-inducible factor 1 (HIF-1) is worth considering in the context of adaptation to IH. We've already shown that patients with OSA are characterized by overexpression of HIF-1 [65][66][67], with other groups reporting similar results [68]. HIF-1 is a factor that mediates hypoxia-dependent response. Among the numerous targets, BDNF is one of them [69], and potentially overexpression of HIF-1 may affect the BDNF signaling pathway in patients with OSA, but it needs further research.
One of the main findings of our study is that patients with OSA with poor sleep and insomnia have higher evening BDNF and proBDNF levels. Moreover, the level of both proteins was positively correlated with total sleep time (TST). Even though the effect sizes of these differences in our studies were low, they were in line with Kaminska et al. and More et al., who showed a similar association between daytime sleepiness and BDNF overexpression in OSA [32,70]. In these studies, daytime sleepiness was evaluated by ESS. Yet, the above outcomes contradict the recent results of insomnia studies, where objective sleep and subjective sleep were correlated with lower serum levels of BDNF [71][72][73]. Fan et al. pointed out that decreased BDNF levels characterized insomnia patients with SSD, lower than 6 h compared to insomnia patients with sleep duration ≥ 6 h and controls. Moreover, the SSD group showed impaired neurocognitive functions, which correlated positively with BDNF levels [71]. What is more, Mikoteit et al. investigated subjective insomnia using Insomnia Severity Index (ISI); the severity of symptoms was correlated with decreased BDNF levels. In contrast to Fan's outcomes, Micoteit et al. showed a correlation between decreased BDNF levels and decreased REM sleep in objective insomnia patients, not with sleep duration [72]. Another study on individuals with insomnia also confirmed the relationship between the severity of subjective sleep impairment and lower serum BDNF levels [73]. Down-regulation of BDNF in insomnia could be explained by hyperactivity of the stress response system and inflammation. Hyperactivity of the hypothalamus-pituitary-adrenal glands axis (HPA) is caused by hyperarousal. In insomnia patients, morning cortisol level is increased [74], and nocturnal melatonin production is diminished [75], which can disturb sleep and its structure, for instance, by REM-sleep changes. Claro et al. emphasized the impact of the stress itself on inflammation and BDNF levels without binding them directly [76]. Thereupon, a positive correlation between proBDNF, BDNF, and total sleep time in OSA and a higher level of expression of studied proteins in OSA individuals with poor sleep quality and insomnia symptoms indicate that patients afflicted with this disease might develop specific compensational mechanisms which are associated with increased expression of studied neurotrophins. Another finding, namely a positive correlation between the level of BDNF and its precursor, which could suggest relatively high activity of the BDNF pathway, might partially corroborate this hypothesis. A possible adaptive mechanism of BDNF pathways upregulation in OSA was described above. This is in line with results form a study where groups did not differ regarding age and sex [33]. In conclusion, BDNF levels may be useful in defining the insomnia phenotype in patients with OSA characterized by excessive daytime sleepiness. It is necessary to understand the complexity of the relationship between sleep disturbances and BDNF. Further research on the exact mechanisms driving this association is warranted.
Circadian clock disruption is a feature property of OSA. Our recent studies confirmed decreased levels of circadian clock proteins, such as period 1 protein (PER1) and aryl hydrocarbon receptor nuclear translocator-like protein 1 (BMAL1), in patients with OSA [77,78]. BDNF protein's expression and its mRNA show dependence on the circadian rhythm: mRNA and BDNF protein levels are elevated during biological night and day, respectively [79]. In this study, we found higher BDNF and proBDNF levels in the morning than in the evening only in the AIS (−) and the PSQI (−) groups. The lack of significant differences in the protein levels in the AIS (+) and the PSQI (+) groups may indicate a possible connection with impaired circadian rhythm.
OSA is well known for neurobehavioral and cognitive deficits, such as decreased attention and vigilance, phonological problems, irritability, and impairment in executive functions and the long-term memory [26]. Neurocognitive impairment was correlated with decreased BDNF levels in the hippocampus several times in mice models of OSA [49,53,79]. Its pathomechanism is complex; nevertheless, it is based on the IH. Oxidative stress caused by IH damages synapses and neurofilaments directly, including postsynaptic density protein 95 failure, impairs new synaptic connections' development [49], and inhibits serotonergic signaling. Wall et al. found that mice had impaired hippocampal neuroplasticity after seven days of IH, measured by long-term potentiation (LTP) in the CA1 region but not in the dentate gyrus [80]. Similar outcomes were obtained by Xie et al. [53]. In the latter study's proposed mechanism, IH and sleep fragmentation directly lead to decreased neuronal excitability, decreased BDNF expression, and enhanced generation of reactive oxygen species [26]. They all impact neurocognitive dysfunction by impairing synaptic plasticity and promoting neuronal apoptosis, which contributes to depression. This could be confirmed to some extent by decreased morning BDNF and proBDNF levels with a medium effect size in the OSA group, with BDI (+) scores observed in our study. It would mean that reduction of BDNF in patients with OSA with depression symptoms may further exacerbate symptoms of this disease and hinder the therapeutic effects of antidepressants.
The main limitation of the study was the small size of the groups. Moreover, insomnia and depression assessments were based on questionnaires, with no other clinical investigation. Additionally, the study design (cross-sectional) prevents a conclusion on the causality. Prospective, interventional trials are necessary to understand the nature of this relationship.

Conclusions
Results of the study suggest that BDNF and proBDNF may be associated with symptoms of insomnia and depression in patients with OSA. Higher BDNF levels may define OSA phenotypes with comorbid insomnia, provide a better description of this heterogenic disorder, and further support proper treatment decisions. Similarly, lower BDNF and proBDNF levels may define OSA phenotypes with intensified symptoms of depression. However, the lack of differences in those protein levels between OSA and control groups indicates the greater complexity of the relationship. Further research is needed to verify the role of BDNF and its precursor in the development of psychiatric OSA comorbidities and assess the effect of treatment on these proteins. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data will be made available upon request.

Conflicts of Interest:
The authors declare no conflict of interest.