Journal ofAffective Disorders, 26 (1992) 191-198
0 1992 Elsevier Science Publishers B.V. All rights reserved 01650327/‘92/$05.00
I91
JAD OO945
aimon, Naoto Yamada, Tetsushi Tsujimoto and Saburo Takahashi Department of Psychiatry, Shiga University of Medical Science, Seto Tsukinowa-cho, Otsu, Shiga Japan
(Received 18 March 1992) (Revision received 24 July 1992) (Accepted 7 August 1992)
Summary We investigated circadian rhythms of body temperature in 62 inpatients with major depressive episodes, by monitorina the deep body temperat!- 3 through the abdominal skin every two hours for a consecutive 48-h period. The data were analyzed by both the least-squares method and the maximum entropy spectral analysis (MEM) and were compared with those in 29 normal volunteers who apparently had a regular 24-h sleep-wake schedule. Circadian rhythm phase disturbances in the depressed patients were likely to be manifested in a phase normal or a phase delay pattern rather than in a phase advance pattern. The amplitude of body temperature was significantly smaller and the mesor was higher in the depressed patients than in the normal subjects. Analysis by MEM revealed that the periods of circadian rhythm of body temperature tended to be longer in the depressed patients than in the normal subjects, though there was no significant difference. The power spectral density by MEM was significantly lower, and there were significantly more ultradian rhythm components in the depressed patients than in the normal subjects. These findings suggest that the fundamental rhythm disturbance in depression may be a weakening of thi: coupling processes between internal pacemakers and an abnormal sensitivity to environmental information.
Key WOT& Depression;
Circadian rhythm; Ultradian
In recent years, it has been noted that patients with affective disorders exhibit abnormalities in
Correspondence
to: K. Daimon, Department of Psychiatry, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu, Shiga 520-21, Japan.
rhythm; Deep body temperature
circadian rhythms of physiological variables. Alterations in circadian rhythms have been reported on sleep (Kupfer, 1976; Wever, 1979; Moline et al., 19&S), plasma and urinary cortisol concentrations (Kathol and Gehris, 1986), plasma prolactin concentrations (Mendlewicz et al., 19801, ply-ma melatonin concentrations (Hariharasubramanian et al., 1986), electrolyte excretion (kobban ct al., 1963; Moody and Allsopp, 1969), motor activity
(Wehr et al., 1982; Wolff et al., 1985; RoyantParola et al., 19861, and body temperature (Kripke et al., 1979; Dirlich et al., 1981: Avery et al., 1982). The depressed patients are also characterized by some features such as early morning awakening and diurnal mood swing which may reflect their physiological functions desynchronizing with environmental rhythms. Based on these previous findings, several hypothetical models involving circadian rhythm dysfunction have been presented for the etiology of affective disorders: (1) phase advance (Wehr and Goodwin, 1981; Wehr and Wirz-Justice, 1982); (21 desynchronization of rhythms (Halberg, 1968); and (3) other abnormalities of the circadian rhythm. Our research group established a method to analyze circadian body temperature pattern by measuring the deep body temperature using a newly developed device (Tsujimoto et al., 1990a). The data obtained were processed by the leastsquares method (COSINORI and tb’ maximum entropy spectral analysis (MEM) (Akaike, 1969: Chen and Stegen, 1974), and we reported that the essential features of the circadian rhythm abnormalities in depression were not exclusively an advanced phase shift but variable and instable features of the circadian rhythm (Tsujimoto et al., 1990b). In the present study, we measured the body temperature in 6, inpatients with major depressive disorders and 29 normal controls, and investigate m using new analyses including power spectral density, ultradian analysis, period analysis, and the relationship between phase and amplitude, in order to elucidate the disturbance of the circadian rhythms in depression in further detail.
TABLE
1
Background
Sixty-two patients with major depressive disorders, aged 25-63 years old, and 29 normal controls, aged 25-64 years old, participated in the study (Table 1). Because of insufficient recruitment from normal volunteers, their sex and age were not matched with those of patients. Severity of the depressive episodes at the time of circadian rhythm evaluations was typically mild or
Sex
Age
Depression Normal
(62)
controls (29)
Male
Female
50.0 * 14.4
20
42
13.9
17
12
37.7+
HRSD
18.85-4.X
moderate with Hamilton Rating Sca!e for Depression (HRSD) scores, 18.8 + 4.8 (mean + SD). All patients were admitted to the Psychiatry Ward of the Shiga University of Medical Science Hospital during the study period of 5 years, 1986-1991. Psychiatric diagnoses were made according to -III criteria, or DSM-III-R criteria after it was revised in 1987. The patients consisted of those with major depression ( N = 44), or bipolar disorder, depressed (N = 18). All patients were in nonseasonal major depressive episodes. All the femaie patients and control subjects were in the lutea! phase of their menstrual cycle or postmenopausal during the measurement of body temperature. Drugs prescribed are listed in Table 2. From the clinical point of view, it was not possible to keep all the depressed patients drugfree during the rhythm assessment. Half of the patients was on various antidepressants, because we had no therapeutic reasons to discontinue medications. Prior to the study, each patient received thoroughly medical and neurological examinations, including routine laboratory analysis, and no patients showed any pathological signs. The patients were informed arid gave consent to participate in the study.
TABLE
Subjects
data of the subjects
2
Drug status of the patients Antidepressants lmipramine
IS
Chlomipramine
9
Amitriptyline
4
Sulpiride
2
Lithium Other Drug-free
carbonate
2 3 27
193
Monitoring schedule In the next week following admission, deep body temperature was monitored from 1O:OOh on the first day to 08: 00 h on the third day, using a Core-Temp (Terumo Co. Ltd., Tokyo), the features of which were described elsewhere (Tsujimoto et al., 1990a). Briefly, the device consists of a automatic recorder and a sensor-heater unit with therms-statt which can monitor the deep body temperature continuously by heating up the unit to the surface temperature theoretically the same as that at the depth of 10 cm in the body. We also described elsewhere the reason we selected the schedule of recording every 2 h for 2 consecutive days (Tsujindoto et al., 199Oa). Briefly, this schedule was proved to be adequate for the computer program by COSINOR to calculate parameters of circadian rhythm. The patients were instructed to go to bed at 21: 00 h and to get up at 06: 00 h according to the ward schedule. During the night the hourly sleep status was recorded by the ward staff. The normal controls were allowed to follow their own usual schedules, but instructed to keep their sleep diaries. They recorded their body temperature at home or during working hours on the fixed schedule described above for the patients. Hamilton Rating Scale for Depression (HRSD) of each patient were also assessed on the first day of body temperature monitoring. Data processing Data recorded were computed by two different programs, COSINOR and MEM. From COSINOR, three parameters of circadian rhythm, ihe acrophase (time of the sinusoid peak), amplitude (height of the sinusoid peak above the mean level), and mesor (mean level of the sinusoid) were calculated. Circadian period and other ultradian rhythm components were calculated by MEM. This method can produce high resolution spectra with accurate frequency and power estimates even for short data sequences (Radoski et al., 1975; Ulrych, 1972, 1975). Because our patients with depression were forced to follow the ward routine schedule, while our normal subjects recorded their body temperatures according to their own individual daily schedules, the timing of sleep in the latter should be adjusted to the
former in calculation. For the purpose, we identified the mid-point of sleep period as circadian time 0: 00 (CT 0). All times described in the paper are indicated in CT. The mean of the mid-point was 1: 30 a.m. in the depressives and 3 : 17 a.m. in the controls. For statistical analyses, analysis of variance (ANOVA), chi-square, and Student’s t-test were used.
Acrophase Normal controls showed a normal distribution pattern of the acrophases with the average of CT 14: 24 + 84 min (SD). On the basis of the result, we defined the normal acrophase range as between CT 11: 36 and CT 17 : 12 (mean - 2SD and mean + 2SD). Forty-four of the 62 depressed subjects (71%) had their acrophases within the normal range, while four patients (6%) had advanced and fourteen patients (23%) had delayed acrophases. Individuals whose acrophases were out of the normal range were 18 of 62 in total, and this proportion did not reach the statistical significance as compared to the normal controls (2 of 29). Amplitude Amplitudes of the body temperature in the depressed group (0.32 + 0.14) (Fig. 1: bottom) were significantly smaller than the value of the control group (0.44 + 0.13) (Fig. 1, top) (P < 0.001; t = 3.47). Amplitudes of the depressed patients whose acrophases stayed in the normal range (0.35 * 0.12) were also significantly smaller than those of the 27 control subjects whose acrophases stayed in the normal range (0.45 + 0.13) (P < 0.005; t = 2.97). Moreover, amplitudes of the depressed patients whose acrophases were outside of the normal range (0.24 + 0.14) were significantly smaller than those of the patients whose acrophases were within the normal range ( P < 0.005; t = 3.01). Mesor Mesors of the depressed be higher than those of though the average of the 0.39) (Fig. 2, bottom) did
patients were likely to the normal controls, former group (36.55 k not suffice the signifi-
194
cant difference from the latter (36.43 + 0.32) (Fig. 2, top) (P < 0.10; t = 1.50). However, mesors of the Jepressed patients whose acrophases were within the normal range (36.62 + 0.34) were significantly higher than those of the 27 control subjects whose dcrophases were within the normal range (36.42 -t_0.33) (P < 0.05; t = 2.27). Moreover, in the depressed patients, the mesors of patients whose acrophases were within the normal range were significantly higher than those
‘C’
Mesor
-
38;
1 I
351
/
/
al
, 0
I
__200
400
A
Acrophase
600
800
@iin)
Amplitude 09
“
0 1992 Elsevier Science Publishers B.V. All rights reserved 01650327/‘92/$05.00
I91
JAD OO945
aimon, Naoto Yamada, Tetsushi Tsujimoto and Saburo Takahashi Department of Psychiatry, Shiga University of Medical Science, Seto Tsukinowa-cho, Otsu, Shiga Japan
(Received 18 March 1992) (Revision received 24 July 1992) (Accepted 7 August 1992)
Summary We investigated circadian rhythms of body temperature in 62 inpatients with major depressive episodes, by monitorina the deep body temperat!- 3 through the abdominal skin every two hours for a consecutive 48-h period. The data were analyzed by both the least-squares method and the maximum entropy spectral analysis (MEM) and were compared with those in 29 normal volunteers who apparently had a regular 24-h sleep-wake schedule. Circadian rhythm phase disturbances in the depressed patients were likely to be manifested in a phase normal or a phase delay pattern rather than in a phase advance pattern. The amplitude of body temperature was significantly smaller and the mesor was higher in the depressed patients than in the normal subjects. Analysis by MEM revealed that the periods of circadian rhythm of body temperature tended to be longer in the depressed patients than in the normal subjects, though there was no significant difference. The power spectral density by MEM was significantly lower, and there were significantly more ultradian rhythm components in the depressed patients than in the normal subjects. These findings suggest that the fundamental rhythm disturbance in depression may be a weakening of thi: coupling processes between internal pacemakers and an abnormal sensitivity to environmental information.
Key WOT& Depression;
Circadian rhythm; Ultradian
In recent years, it has been noted that patients with affective disorders exhibit abnormalities in
Correspondence
to: K. Daimon, Department of Psychiatry, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu, Shiga 520-21, Japan.
rhythm; Deep body temperature
circadian rhythms of physiological variables. Alterations in circadian rhythms have been reported on sleep (Kupfer, 1976; Wever, 1979; Moline et al., 19&S), plasma and urinary cortisol concentrations (Kathol and Gehris, 1986), plasma prolactin concentrations (Mendlewicz et al., 19801, ply-ma melatonin concentrations (Hariharasubramanian et al., 1986), electrolyte excretion (kobban ct al., 1963; Moody and Allsopp, 1969), motor activity
(Wehr et al., 1982; Wolff et al., 1985; RoyantParola et al., 19861, and body temperature (Kripke et al., 1979; Dirlich et al., 1981: Avery et al., 1982). The depressed patients are also characterized by some features such as early morning awakening and diurnal mood swing which may reflect their physiological functions desynchronizing with environmental rhythms. Based on these previous findings, several hypothetical models involving circadian rhythm dysfunction have been presented for the etiology of affective disorders: (1) phase advance (Wehr and Goodwin, 1981; Wehr and Wirz-Justice, 1982); (21 desynchronization of rhythms (Halberg, 1968); and (3) other abnormalities of the circadian rhythm. Our research group established a method to analyze circadian body temperature pattern by measuring the deep body temperature using a newly developed device (Tsujimoto et al., 1990a). The data obtained were processed by the leastsquares method (COSINORI and tb’ maximum entropy spectral analysis (MEM) (Akaike, 1969: Chen and Stegen, 1974), and we reported that the essential features of the circadian rhythm abnormalities in depression were not exclusively an advanced phase shift but variable and instable features of the circadian rhythm (Tsujimoto et al., 1990b). In the present study, we measured the body temperature in 6, inpatients with major depressive disorders and 29 normal controls, and investigate m using new analyses including power spectral density, ultradian analysis, period analysis, and the relationship between phase and amplitude, in order to elucidate the disturbance of the circadian rhythms in depression in further detail.
TABLE
1
Background
Sixty-two patients with major depressive disorders, aged 25-63 years old, and 29 normal controls, aged 25-64 years old, participated in the study (Table 1). Because of insufficient recruitment from normal volunteers, their sex and age were not matched with those of patients. Severity of the depressive episodes at the time of circadian rhythm evaluations was typically mild or
Sex
Age
Depression Normal
(62)
controls (29)
Male
Female
50.0 * 14.4
20
42
13.9
17
12
37.7+
HRSD
18.85-4.X
moderate with Hamilton Rating Sca!e for Depression (HRSD) scores, 18.8 + 4.8 (mean + SD). All patients were admitted to the Psychiatry Ward of the Shiga University of Medical Science Hospital during the study period of 5 years, 1986-1991. Psychiatric diagnoses were made according to -III criteria, or DSM-III-R criteria after it was revised in 1987. The patients consisted of those with major depression ( N = 44), or bipolar disorder, depressed (N = 18). All patients were in nonseasonal major depressive episodes. All the femaie patients and control subjects were in the lutea! phase of their menstrual cycle or postmenopausal during the measurement of body temperature. Drugs prescribed are listed in Table 2. From the clinical point of view, it was not possible to keep all the depressed patients drugfree during the rhythm assessment. Half of the patients was on various antidepressants, because we had no therapeutic reasons to discontinue medications. Prior to the study, each patient received thoroughly medical and neurological examinations, including routine laboratory analysis, and no patients showed any pathological signs. The patients were informed arid gave consent to participate in the study.
TABLE
Subjects
data of the subjects
2
Drug status of the patients Antidepressants lmipramine
IS
Chlomipramine
9
Amitriptyline
4
Sulpiride
2
Lithium Other Drug-free
carbonate
2 3 27
193
Monitoring schedule In the next week following admission, deep body temperature was monitored from 1O:OOh on the first day to 08: 00 h on the third day, using a Core-Temp (Terumo Co. Ltd., Tokyo), the features of which were described elsewhere (Tsujimoto et al., 1990a). Briefly, the device consists of a automatic recorder and a sensor-heater unit with therms-statt which can monitor the deep body temperature continuously by heating up the unit to the surface temperature theoretically the same as that at the depth of 10 cm in the body. We also described elsewhere the reason we selected the schedule of recording every 2 h for 2 consecutive days (Tsujindoto et al., 199Oa). Briefly, this schedule was proved to be adequate for the computer program by COSINOR to calculate parameters of circadian rhythm. The patients were instructed to go to bed at 21: 00 h and to get up at 06: 00 h according to the ward schedule. During the night the hourly sleep status was recorded by the ward staff. The normal controls were allowed to follow their own usual schedules, but instructed to keep their sleep diaries. They recorded their body temperature at home or during working hours on the fixed schedule described above for the patients. Hamilton Rating Scale for Depression (HRSD) of each patient were also assessed on the first day of body temperature monitoring. Data processing Data recorded were computed by two different programs, COSINOR and MEM. From COSINOR, three parameters of circadian rhythm, ihe acrophase (time of the sinusoid peak), amplitude (height of the sinusoid peak above the mean level), and mesor (mean level of the sinusoid) were calculated. Circadian period and other ultradian rhythm components were calculated by MEM. This method can produce high resolution spectra with accurate frequency and power estimates even for short data sequences (Radoski et al., 1975; Ulrych, 1972, 1975). Because our patients with depression were forced to follow the ward routine schedule, while our normal subjects recorded their body temperatures according to their own individual daily schedules, the timing of sleep in the latter should be adjusted to the
former in calculation. For the purpose, we identified the mid-point of sleep period as circadian time 0: 00 (CT 0). All times described in the paper are indicated in CT. The mean of the mid-point was 1: 30 a.m. in the depressives and 3 : 17 a.m. in the controls. For statistical analyses, analysis of variance (ANOVA), chi-square, and Student’s t-test were used.
Acrophase Normal controls showed a normal distribution pattern of the acrophases with the average of CT 14: 24 + 84 min (SD). On the basis of the result, we defined the normal acrophase range as between CT 11: 36 and CT 17 : 12 (mean - 2SD and mean + 2SD). Forty-four of the 62 depressed subjects (71%) had their acrophases within the normal range, while four patients (6%) had advanced and fourteen patients (23%) had delayed acrophases. Individuals whose acrophases were out of the normal range were 18 of 62 in total, and this proportion did not reach the statistical significance as compared to the normal controls (2 of 29). Amplitude Amplitudes of the body temperature in the depressed group (0.32 + 0.14) (Fig. 1: bottom) were significantly smaller than the value of the control group (0.44 + 0.13) (Fig. 1, top) (P < 0.001; t = 3.47). Amplitudes of the depressed patients whose acrophases stayed in the normal range (0.35 * 0.12) were also significantly smaller than those of the 27 control subjects whose acrophases stayed in the normal range (0.45 + 0.13) (P < 0.005; t = 2.97). Moreover, amplitudes of the depressed patients whose acrophases were outside of the normal range (0.24 + 0.14) were significantly smaller than those of the patients whose acrophases were within the normal range ( P < 0.005; t = 3.01). Mesor Mesors of the depressed be higher than those of though the average of the 0.39) (Fig. 2, bottom) did
patients were likely to the normal controls, former group (36.55 k not suffice the signifi-
194
cant difference from the latter (36.43 + 0.32) (Fig. 2, top) (P < 0.10; t = 1.50). However, mesors of the Jepressed patients whose acrophases were within the normal range (36.62 + 0.34) were significantly higher than those of the 27 control subjects whose dcrophases were within the normal range (36.42 -t_0.33) (P < 0.05; t = 2.27). Moreover, in the depressed patients, the mesors of patients whose acrophases were within the normal range were significantly higher than those
‘C’
Mesor
-
38;
1 I
351
/
/
al
, 0
I
__200
400
A
Acrophase
600
800
@iin)
Amplitude 09
“
