Authors: Etsuko Kita, Natsuki Nakayama, Momoka Niihara, Yoshimi Moriwaki, Aoi Kono, Souta Ookabe, Chiyo Iwata, Misao Kurita
Categories: Research, Autonomic nervous system, Major depressive disorder, Hypersomnia, Heart rate variability, Quick inventory of depressive Symptomatology-Self-Report, Circadian rhythm
Source: BioPsychoSocial Medicine
Authors: Etsuko Kita, Natsuki Nakayama, Momoka Niihara, Yoshimi Moriwaki, Aoi Kono, Souta Ookabe, Chiyo Iwata, Misao Kurita
Hypersomnia, defined as a total sleep time of over 10 h within a 24-h period, is common in patients with major depressive disorder (MDD). The circadian rhythm of patients with hypersomnia and depression is disturbed; however, differences from patients without hypersomnia remain unclear. We aimed to clarify differences over 24 h in the autonomic nervous system (ANS) activity of patient groups with MDD with or without hypersomnia.
This study included outpatients with MDD who were categorized into either a hypersomnia or non-hypersomnia group based on the Diagnostic and Statistical Manual of Mental Disorders-5 criteria. Heart rate variability data were collected over 24 h using electrocardiograms. Differences in ANS activity between groups across bedtime and non-bedtime periods were analyzed using two-way ANOVA. General linear models were used to compare hourly 24-h data, with heart rate; natural logarithm (ln) of the low frequency domain (LF) as sympathetic and parasympathetic nerve activity; ln of the high frequency domain (HF) as parasympathetic nerve activity; ln LF/HF as objective variables indicating sympathetic domination; hypersomnia status every hour during the 24-h period as explanatory variables; and age and body mass index as variable effects.
Twenty-eight participants were enrolled; the data of 25 were available for the final analysis. No interaction between hypersomnia status and period was observed for ANS indices. ln LF differed significantly by hypersomnia status but not by period (bedtime vs. non-bedtime); the hypersomnia group showed higher ln LF across both periods. In the evening, ln HF was significantly higher in the hypersomnia group than in the non-hypersomnia group, whereas ln LF/HF was significantly higher in the early morning in the hypersomnia group.
No interaction between hypersomnia status and period (bedtime and non-bedtime) was observed for autonomic nervous activity indices. However, in the hypersomnia group, hourly analyses showed significantly higher ln HF from 00 to 00 h and higher ln LF/HF before waking (3:00–4:00 h) compared with the non-hypersomnia group. ANS activity differed depending on presence or absence of hypersomnia, a risk factor for relapse and bipolar disorder.
Mood disorders, such as unipolar major depressive disorder (MDD) and bipolar disorder, are major global public health problems [1]. Hypersomnia is defined as a total sleep time of over 10 h within a 24-hour period and is common in patients with mood disorders, including MDD [2]. In this study, “hypersomnia” refers to a DSM-5 depressive disorder symptom [2], not to the “hypersomnia/hypersomnolence” identified by the International Classification of Sleep Disorders, Third Edition (ICSD3). In DSM-5 [3], hypersomnia in depression is defined as a symptom and is assessed with a single question about whether the patient sleeps too much [3]. By contrast, hypersomnia classified by ICSD3 includes narcolepsy types I and II, idiopathic hypersomnia, and Kleine-Levin syndrome, which are diagnosed using objective measures such as Polysomnography (PSG) and multiple sleep latency testing. However, prior research has noted that there is no strict boundary between the “hypersomnia” described in DSM-5 and the hypersomnia disorder defined in ICSD-3 [4]. Narcolepsy has been associated with a high level of self-reported depressive symptoms, and links between sleep disorders and depressive symptoms have been reported [3]. Silvani et al. reported that hypersomnia disorders as defined in ICSD-3 are associated with ANS dysfunction [5], and Sfonza et al. reported ANS dysfunction in narcolepsy [6]. Idiopathic hypersomnia shows delayed-phase secretion of melatonin and cortisol, indicating a circadian rhythm disturbance [7]. Accordingly, depressive patients with hypersomnia may have impaired autonomic function and circadian rhythm regulation.
Hypersomnia is a predictor of the prognosis of patients with MDD [8]. Furthermore, it is associated with treatment resistance, symptomatic relapses, increased risk of suicide, and functional impairment [8]. The prevalence of hypersomnia was reported to be lower than that of insomnia in patients with major depressive episodes [9, 10]. Residual symptoms worsen the prognosis of patients with MDD. In a study of 1,133 outpatients with non-psychotic MDD, the hazard ratio for hypersomnia was 1.19 after achieving remission with level 1 treatment (citalopram for up to 14 weeks) [11]. However, the risk of bipolar disorder in patients with comorbid hypersomnia and insomnia increased by two- to three-fold [9, 11]. Sleeping for over 10 h in a 24-hour period is a common symptom of outpatients with MDD, regardless of whether their depressive state is mild.
Melancholic depression and atypical depression are two distinct subtypes of MDD. Hypersomnia is a typical symptom of atypical depression, which also presents mood reactivity to positive events and at least two of the following increased appetite or increased weight, hypersomnia, leaden paresis, and chronic rejection sensitivity [2]. In addition, the depressive symptoms are worse in the evening [12]. In contrast, melancholic depression presents with insomnia and reduced appetite, and the depressive symptoms are worse in the morning [12]. Therefore, the patterns of depressive symptoms differ between patients with hypersomnia and those without.
The alteration of sleep-awake patterns, melatonin and cortisol secretion patterns, and changes in body temperature are the hallmarks of MDD [13]. Notably, depression is caused by abnormal regulation of the hypothalamic–pituitary–adrenal (HPA) axis and activation of inflammatory responses [14, 15]. Corticotropin-releasing hormone (CRH) is released from the hypothalamus in response to stress and causes the secretion of adrenocorticotropic hormone (ACTH) from the anterior pituitary gland. This causes cortisol release from the adrenal cortex into the blood [16]. Increased cortisol levels suppress the secretion of CRH and ACTH through a negative feedback mechanism. Patients with melancholic depression exhibit HPA axis dysfunction, including excessive ACTH and cortisol secretion and non-suppression of cortisol secretion after dexamethasone administration [15, 17, 18]. Reportedly, CRH is reduced in the central nervous system of patients with atypical depression. In addition, HPA axis activity is reportedly reduced in atypical depression, compared to melancholic depression [19, 20]. Normally, HPA axis activation occurs under the influence of non-associated stress associated with circadian rhythm control, in which the highest cortisol levels are observed early in the morning [14, 16]. Cortisol secretion increases early in the morning in melancholic depression, like healthy individuals, and it peaks around 00, but the level of secretion is high relative to that of healthy individuals [21]. Stewart et al. reported that the regulation of cortisol secretion differed between atypical and melancholic depression [22]. Therefore, we consider that the circadian rhythm in atypical depression differs from that of melancholic depression.
Because hypersomnia is a characteristic of atypical depression with low cortisol levels, which affect sleep-awake patterns, the circadian rhythm of patients with hypersomnia and MDD might differ from those with non-hypersomnia with MDD. Dysfunctional circadian rhythms have also been shown to be causally involved in the pathogenesis of bipolar disorder [23]. Because the risk of patients with hypersomnia and MDD developing bipolar disorder is high [9], we consider that the circadian rhythm differs in patients with or without hypersomnia.
Heart rate variability (HRV) is a useful method for non-invasively comprehending the circadian rhythm of outpatients [24]. Previous research has been conducted on the circadian rhythm and autonomic nervous system (ANS) activity of patients with MDD [25–27]. Like healthy individuals, the parasympathetic nervous system (PNS) index shows two levels over 24 h in patients with depression, one high and one low; however, the transition has been shown to take approximately 10 h [28]. The PNS index, which is predominantly high during sleep, becomes low within 1 h of waking in healthy individuals [28]. Prior studies have shown differences in ANS activity between patients with MDD and healthy individuals. Hypersomnia is one symptom of atypical depression, with circadian rhythm disturbances also reported [19, 20]. Because hypersomnia in depression may share features with ICSD3 sleep disorders and is a symptom of atypical depression, autonomic nervous function may differ even among patients with MDD. However, whether ANS circadian rhythms differ between patients with MDD who have hypersomnia and those who do not remains unclear.
To determine the difference in the ANS activity and circadian rhythm of patients with or without hypersomnia, it is useful to discuss physiological risk factors that are associated with developing bipolar disorder or recurrent depression. Therefore, in this study, we aimed to identify differences in ANS activity during a 24-h period between groups of patients with MDD with and without hypersomnia.
Data was collected at the outpatient department of one psychiatric hospital and two psychiatric clinics from February to August 2021. We included patients who (1) were aged > 20 years, (2) visited a doctor regularly at the outpatient departments of these institutions, (3) met the diagnostic criteria for MDD, and (4) obtained approval from their doctor about participating in this study. We excluded postoperative patients, those with complications from any progressive physical disease, those receiving pacemaker implantation, and those requiring inpatient treatment. The participants who met the study eligibility criteria were recruited by having their attending physicians introduce them to the researchers.
We recorded basic participant characteristics (age, sex, height, weight), psychological history, age of onset, and time from onset of depression. Data on medical history was obtained from medical records at the respective institutions. Subjective depressive symptoms were investigated using the 16-item Quick Inventory of Depressive Symptomatology-Self-Report (QIDS-SR) [29]. This scale consists of nine categories and 16 items and is used for clinical evaluation of the severity of depression. The Japanese version of QIDS-SR was used for this study [30]. Participants answered 16 questions in four categories. These included four questions each on sleep and appetite, two on psychomotor self-evaluation, and one question each on low mood, impaired concentration/decision-making, negative self-view, suicidal ideation, lack of involvement, and loss of energy. A score of 0–3 points was assigned for each question. The score for all other items was directly included. Only the highest score for the questions on sleep, appetite, and psychomotor self-evaluation were recorded. The total scores ranged from 0 to 27 points. A score of 0–5 was considered normal, 6–10 mild, 11–15 moderate, 16–20 severe, and 21–27 extremely severe.
HRV data served as an indicator of ANS activity. The study participants were fitted with a Holter electrocardiograph (FM-960, Holter recorder, Fukuda Denshi, Tokyo, Japan) for 24-h and electrocardiograms were obtained. We divided the electrocardiogram data into 5-min intervals, expressed the magnitude of variation in the RR interval as a function of frequency, and performed frequency analysis to quantify variables by dividing them into different components at different frequencies. We collected heart rate (HR) data in the low (LF, 0.04–0.15 Hz) and high frequency domains (HF, 0.15–0.4 Hz) from the electrocardiogram data [31]. HRV is the variation in the RR interval of heartbeats and reflects the activity of the efferent sympathetic and vagus nerves [31]. The normal RR interval recorded on the electrocardiogram is due to the depolarization of the sinus node. LF reflects sympathetic and parasympathetic nerve activity. HF reflects parasympathetic or vagal nerve activity, whereas a higher LF/HF ratio reflects sympathetic domination [31, 32]. The average ANS activity index was calculated from the 24-h data on increases and decreases in HR and individual self-reports. We defined “bedtime” as the period from going to bed at night to waking the next day and “non-bedtime” as the rest of the day after waking.
For the question on hypersomnia in the QIDS-SR, the participants who answered “0: no problem” were placed in the non-hypersomnia group. Those who answered “1: about 10 h of sleep, including nap time within a 24-h period,” or “2: about 12 h of sleep, including nap time within a 24-h period,” or “3: >12 h of sleep, including nap time within a 24-h period,” were placed in the hypersomnia group. Group attributes, including body mass index (BMI), QIDS-SR score, and clinical characteristics, were calculated.
A t-test was conducted to determine if the hypersomnia and non-hypersomnia groups had different attributes. We calculated the average HRV index during bedtime and non-bedtime for each group and each period. The average values of HR and ANS indicators (LF, HF, LH/HF) were calculated hourly and during bedtime and non-bedtime for each group. The normality of the bedtime and non-bedtime ANS indices and hourly values for each group were confirmed using the Shapiro-Wilk test. Normality in each group was observed only in HR during bedtime and non-bedtime: the other ANS indicators were not, although there were some exceptions, thus, they were logarithmically transformed to evaluate the differences in ANS activity. Between group differences in ANS activity across bedtime and non-bedtime periods were analyzed using two-way ANOVA, with age and body mass index (BMI) included as covariates. A general linear model was used to compare continuous 24-h data between groups, with HR, natural logarithm (ln) of LF, ln HF, and ln LF/HF as objective variables, hypersomnia status and every hour during the 24-h as explanatory variables, and age and BMI as variable effects. Multiple comparisons were not adjusted for in this exploratory data analysis. The IBM Statistical Package for Social Sciences (Ver. 29) was used for analysis. The statistical significance level was set at p <.05.
This study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the ethics review committee of the researchers’ institution (approval 20–129). This observational, exploratory study did not involve any prospective intervention; therefore, clinical trial registration was not required under the ICMJE definition. The study protocol was reviewed and approved by the institutional review board, and written informed consent was obtained from all participants.
Twenty-eight participants were enrolled and three were excluded due to missing HRV data, leaving the data of 25 available for analysis. No participants were receiving medical treatment for cardiovascular disease. Nine participants had hypersomnia and 16 did not. Of the nine participants with hypersomnia, six had sleep disturbances, difficulty falling asleep, arousal during sleep, and episodes of early-morning awakening. The hypersomnia group included five men (55.6%): average age 50.44 years (standard deviation [SD] 11.65), average time since MDD diagnosis nine years (SD 9.08), and average BMI 25.65 (SD 5.63). The non-hypersomnia group included 10 men (62.5%): average age 52.31 years (SD 10.31), average time since MDD diagnosis 10.25 years (SD 5.73), and average BMI 28.30 (SD 5.56). The average QIDS-SR scores were 8.78 (SD 3.80) and 7.75 (SD 6.87) in the hypersomnia and non-hypersomnia groups, respectively. No significant differences were observed between the hypersomnia and non-hypersomnia groups regarding average age, age at the time of MDD onset, time since MDD diagnosis, BMI, or QIDS-SR score. All patients had taken medication (antidepressants, antipsychotics, or sedative-hypnotics), but there was no difference between groups in the medication equivalent level. None of the participants in either group were treated with bright light therapy (Table 1).
Table 1Demographic and clinical characteristicsHypersomnia (n = 9)Non-hypersomnia (n = 16)t-valuep-valueAge [years, mean(SD)]50.44 (11.65)52.31 (10.31)−0.4150.682 ^a^Sex [male, n (%)]5 (55.6)10 (62.5)0.734 ^b^Age of onset [years, mean(SD)]41.44 (12.36)42.06 (8.97)−0.1440.886 ^a^Length since diagnosis of MDD [years, mean(SD)]9 (9.08)10.25 (5.73)−0.4240.676 ^a^BMI [mean(SD)]25.65 (5.63)28.30 (5.56)−1.1420.265 ^a^QIDS-SR [mean(SD)]8.78 (3.80)7.75 (6.87)0.4120.684 ^a^ Sleep disturbance [n (%)]9 (100.0)14 (87.5) Difficulty getting to sleep [n (%)]4 (44.4)9 (56.2) Arousal during sleep [n (%)]6 (66.7)10 (62.5) Early-morning awaking [n (%)]4 (44.4)9 (56.2) Hypersomnia [n (%)]9 (100.0)0 (0.0)Non-bedtime length [mean]16 h 18 min16 h 39 minBedtime length [mean]7 h 41 min7 h 20 minSleep-onset time [mean]22:5123:13Arising time [mean]6:326:33Antidepressant medication use [n (%)]8 (88.9)13 (62.5) Tricyclic [n (%)]0 (0.0)2 (12.5) SSRI [n (%)]3 (33.3)6 (37.5) SNRI [n (%)]5 (55.6)6 (37.5) NaSSA [n (%)]1 (11.1)3 (18.8)Imipramine equivalents (mg/day) [mean (SD)]189.06 (111.69)145.19 (93.20)0.9720.343^a^ Sedative-hypnotic medication use [n (%)]6 (66.7)14 (87.5)Diazepam equivalents (mg/day) [mean (SD)]8.95 (6.16)6.99 (5.84)0.6790.506^a^ Antipsychotic medication use [n (%)]4 (44.4)9 (56.2)Chlorpromazine equivalents (mg/day) [mean (SD)]93.75 (37.50)116.67 (135.21)−0.3260.750^a^SD standard deviation, BMI Body Mass Index, QIDS Quick Inventory of Depressive Symptomatology Self-Report, Tricyclic tricyclic antidepressant, SSRI serotonin selective reuptake inhibitor, SNRI serotonin norepinephrine reuptake inhibitor, NaSSA noradrenergic and specific serotonergic antidepressantsa t-test, b Fisher’s exact test
A two-way ANOVA was performed to evaluate the effects of hypersomnia/non-hypersomnia on HR during bedtime/non-bedtime. The means and standard deviations for HR are presented in Table 2.
The results indicated no significant main effect for hypersomnia/non-hypersomnia, F(1, 44) = 2.916, p =.095, partial η^2^ = 0.062; a significant main effect for non-bedtime/bedtime, F(1, 44) = 29.703, p =.000, partial η^2^ = 0.403; and no significant interaction between hypersomnia/non-hypersomnia and bedtime/non-bedtime, F(1,44) = 1.864, p =.179, partial η^2^ = 0.041 (Fig. 1A).
Fig. 1Differences in bedtime and non-bedtime HR, lnLF, lnHF, and lnLF/HF MDD, major depressive disorder. Panels show HR (A), lnLF (B), lnHF (C), and lnLF/HF (D) in the hypersomnia and non-hypersomnia groups. Results were analyzed by two-way ANOVA, adjusted for age and body mass index
Two-way ANOVAs (hypersomnia/non-hypersomnia × bedtime/non-bedtime) were performed for ln LF, ln HF, and ln LF/HF; means and standard deviations are shown in Table 2. For ln LF, there was a significant main effect of hypersomnia/non-hypersomnia, F(1, 44) = 4.442, p =.041, partial η^2^ = 0.092, with no main effect of bedtime/non-bedtime,
Table 2Descriptive statistics for ANS activity indicesGroupStateMeanSDHRhypersomnianon-bedtime76.6328.927bedtime66.1237.727non-hypersomnianon-bedtime87.61112.245bedtime70.0799.882ln LFhypersomnianon-bedtime5.6910.509bedtime6.0070.893non-hypersomnianon-bedtime4.9860.855bedtime5.4820.896ln HFhypersomnianon-bedtime4.6430.595bedtime4.9841.065non-hypersomnianon-bedtime3.7670.842bedtime4.7681.021ln LF/HFhypersomnianon-bedtime1.2380.433bedtime1.2350.541non-hypersomnianon-bedtime1.4670.417bedtime0.8900.644
F (1, 44) = 2.758, p =.104, partial η^2^ =.059, and no interaction, F (1, 44) =.137, p =.713, partial η^2^ =.003 (Fig. 1B). For ln HF, there was no main effect of hypersomnia/non-hypersomnia, F(1, 44) = 2.742, p =.105, partial η^2^ =.059; a significant main effect of bedtime/non-bedtime, F(1, 44) = 6.163, p =.017, partial η^2^ =.123; and no interaction, F(1, 44) = 1.485, p =.229, partial η^2^ =.033 (Fig. 1C). For ln LF/HF, neither main effect was significant—hypersomnia/non-hypersomnia: F(1, 44) =.381, p =.540, partial η^2^ =.009; bedtime/non-bedtime: F(1, 44) = 3.666, p =.062, partial η^2^ =.077—and the interaction was also non-significant, F(1, 44) = 3.594, p =.065, partial η^2^ =.076 (Fig. 1D).
An hourly assessment of HR and ANS activity showed a significant difference in all measured indices. HR was significantly lower in the hypersomnia group than in the non-hypersomnia group at 00–13:00 h, at 00 h, and at 00–19:00. In the analysis of HR, the effect sizes for the main effect of hypersomnia were moderate (partial η^2^ = 0.057), and the main effect of time was large (partial η^2^ = 0.276). However, the interaction between time and hypersomnia was small (partial η^2^ = 0.032). In addition, ln LF was significantly higher in the hypersomnia group relative to the non-hypersomnia group at 00 h and at 00–22:00 h. In the analysis of ln LF, the main effect of hypersomnia was a small effect size (partial η^2^ = 0.034), and the main effect of time was a medium effect size (partial η^2^ = 0.065), but the time and hypersomnia interaction was a small effect size (partial η^2^ = 0.048). Further, ln HF was significantly higher in the hypersomnia group than in the non-hypersomnia group at 00–22:00 h. In the analysis of ln HF, the main effect of hypersomnia was small (partial η^2^ = 0.009), but the main effect of time was medium (partial η^2^ = 0.067), and the interaction between time and hypersomnia was also medium (partial η^2^ = 0.052). In the hypersomnia group, ln LF/HF was significantly higher at 00 and 00 h and significantly lower at 00 h, compared to the non-hypersomnia group. In the analysis of ln LF/HF, the main effect of hypersomnia was partial η^2^ = 0.002, the main effect of time was partial η^2^ = 0.028, and the interaction of time and hypersomnia was partial η^2^ = 0.037, all of which had small effect sizes (Fig. 2).
Fig. 2Differences in the hourly ANS activity of patients with MDD with and without hypersomnia ANS, autonomic nervous system activity. *: p<.05, **: p<.01. HR: Heart Rate, LF: low frequency domain, HF: high frequency domain, LF/HF: low frequency domain/high frequency domain
To our knowledge, this is the first study to show differences in ANS activity during bedtime and non-bedtime periods in patients with MDD with and without hypersomnia, with hypersomnia defined as over 10 h of sleep in a 24-hour period, according to the DSM-5 criteria. Previous studies had used the Pittsburg Sleep Quality Index and Epworth Sleepiness Scale (ESS) to evaluate hypersomnia [33–36]. These studies measured the frequency of sleepiness in daily life, unlike our study, which solely used DSM-5 criteria for identifying hypersomnia. Yang et al. had previously reported that non-bedtime sleepiness was not associated with ANS activity [37]. In their study, hypersomnia was evaluated by measuring non-bedtime sleepiness using the ESS. One possible reason for observing a correlation between hypersomnia and ANS activity was because the present study criterion was hypersomnia.
In our study, no interaction between hypersomnia status and bedtime/non-bedtime was observed for autonomic nervous activity indices. A previous study reported higher HF during sleep than during wakefulness [38]. ln LF differed by hypersomnia status but not by period (bedtime vs. non-bedtime). Regardless of period, the hypersomnia group showed higher ln LF. Because interpretation of LF requires caution [31], the reason for higher values in the hypersomnia group is unclear; nevertheless, this may be a characteristic of depressed patients with hypersomnia. In the non-hypersomnia group, ln LF/HF during bedtime was significantly lower than during non-bedtime, whereas in the hypersomnia group it appeared unchanged; however, the difference was not statistically significant. According to previous reports, bedtime HF is significantly higher than non-bedtime HF, whereas ln LF/HF is lower [24, 39]. Participants in the present study had mild depression. Several studies have reported reduced autonomic nervous activity in patients with depression compared with healthy individuals [27, 40]. One possible reason our study did not fully demonstrate bedtime versus non-bedtime differences by hypersomnia status is that autonomic activity may have been reduced by depression itself, irrespective of hypersomnia status.
Hourly changes showed that ln HF, an indicator of PNS activity, was significantly higher from 00–22:00 h in the hypersomnia group compared to the non-hypersomnia group. This difference can be attributed to the increase in ln HF at 00 h in the hypersomnia group and the decrease in the afternoon ln HF of the non-hypersomnia group. In the non-hypersomnia group, In HF was lowest at 00 h and increased toward sleep-onset time. A previous study reported that the ln HF of patients with depression peaks at 30 h, is lowest at approximately 30 h, then increases again at 30 h [28]. This indicates that ln HF peaked before falling asleep in the hypersomnia group and only decreased slightly upon waking. In addition, it decreased toward evening and increased from evening upwards, as seen in patients with depression.
Furthermore, LF/HF, an indicator of sympathetic activity, was significantly higher before waking, at 00 and 00 h, in the hypersomnia group compared to the non-hypersomnia group. The significant differences can be attributed to the variation in ln LF/HF patterns. In the hypersomnia group, ln LF/HF increased from 00 h and peaked at 00 h. In contrast, the ln LF/HF of the non-hypersomnia group was lowest between 00–03:00 h and rapidly increased from 00 h. There are few studies on the LF/HF ratio of patients with MDD done over a 24-h period. Previous research indicated that the LF/HF of male patients with depression was highest at 30–12:00 h, whereas that of female patients was highest at approximately 00 h [26]. However, the participants included not only patients with MDD but also those in the depressive state of bipolar disorder. Thus, the 24-h LF/HF of patients with MDD was not fully clarified. Hypersomnia is a symptom of atypical depression. Geovanni et al. reported that the cortisol secretion level is lower in atypical depression than in melancholic depression, which lowers the activity of the HPA axis, but no significant difference was observed [41]. Thus, the present study showed that the ln LF/HF in participants with hypersomnia was significantly lower at 00 h than that of those with non-hypersomnia. However, evidence supporting this is unclear because no previous research has reported similar findings. In patients with hypersomnia, there appears to be no change from parasympathetic to sympathetic activity when transitioning from sleep to the waking state, whereas high parasympathetic activity in the evening indicates a disrupted circadian rhythm.
In the present study, ln HF, a measure of parasympathetic activity, decreased from early morning, stayed low during the non-bedtime hours, and increased at night in the non-hypersomnia group. Parasympathetic activity becomes dominant when patients without hypersomnia transition from being awake to a sleeping state. Previous studies have reported that the parasympathetic activity of healthy individuals peaks at 00 or 00 h before waking, with peak levels approximately twice as high as the lowest levels [39, 42]. In healthy individuals, sympathetic activity becomes dominant when transitioning from sleep to being awake; sympathetic activity becomes dominant [24, 28]. Furthermore, ln LF/HF, an indicator of sympathetic activity, becomes rapidly active, increases by approximately 0.4 points within 1 h after 00 or 00 h in patients without hypersomnia, stays active during non-bedtime hours, and reduces at night. HR increases around waking time, shows no dynamic reduction during non-bedtime hours, and decreases from 00 h to midnight. According to a previous study of healthy individuals, LF/HF was observed to peak at high noon, the highest level during non-bedtime hours, and then decrease until midnight [43]. Therefore, the patients in the non-hypersomnia group displayed a pattern of ANS activity like healthy individuals throughout the day. Most effect sizes were small (except for Ln HF), limiting interpretability; nevertheless, the results indicate differences in autonomic nervous activity between individuals with and without hypersomnia.
The circadian rhythm of patients with MDD in the hypersomnia group differed from those in the non-hypersomnia group for the following (1) No interaction between hypersomnia status and period (bedtime and non-bedtime) was observed for autonomic nervous activity indices; (2) parasympathetic activity was significantly higher in the hypersomnia group from 00–22:00 h compared to the non-hypersomnia group; and (3) sympathetic activity peaked at 00 h in the hypersomnia group, which was significantly higher than that in the non-hypersomnia group. Previous research reports that dysregulation or instability of circadian patterns and sleep-wake cycles contribute to the development of bipolar disorder because circadian rhythm is associated with emotion or activity regulation. The mood stabilizer, lithium, affects the genes of the circadian clock, indicating a link between the regulation of circadian rhythm genes and bipolar disorder [23]. Impaired circadian rhythms in patients with hypersomnia might be the cause of bipolar disorder.
This study has some limitations, and the findings should be interpreted with caution regarding generalizability. Because participants were recruited from three clinical settings and assessed under specific measurement conditions, the results are most directly applicable to adults with depressive disorders similar to our cohort. The modest sample size and potential influences of age, sex, medication status, and co-occurring sleep disturbances may further limit broader applicability. Nevertheless, timely evidence on how autonomic activity varies with hypersomnia in depression has clinical implications for understanding pathophysiology, recognizing patients’ primary complaints, informing treatment, improving quality of life, and supporting relapse prevention. All participants received medication, and no between-group differences in medication dose were observed; however, prior research indicates that tricyclic antidepressants can significantly reduce HR [44]; thus, medication may have affected the HRV measurements in this study.
This study included participants with mild depressive symptoms. Studies reporting ANS activity in patients with MDD typically compared them with healthy controls. Similarly, in the present study, clear characterization of the ANS activity of patients with and without hypersomnia may help identify physiological factors that increase the risk of recurrence.
The present study reported differences in ANS activity within a 24-h period in groups of patients with MDD with or without hypersomnia. No interaction between hypersomnia status and period (bedtime and non-bedtime) was observed for autonomic nervous activity. However, hourly changes in the hypersomnia group showed that ln HF was significantly higher at 00–22:00 h and that ln LF/HF was significantly higher before waking, at 00 and 00 h, compared to the non-hypersomnia group. ANS activity depends on whether hypersomnia, a risk factor for relapse and bipolar disorder, is present.