BIOL PSYCHIATRY 1990;28:569--587

569

Neuroendocrinological Investigations During Sleep Deprivation in Depression II. Longitudinal Measurement of Thyrotropin, TH, Cortisol, Prolactin, GH, and LH During Sleep and Sleep Deprivation Andreas Baumgartner, Dieter Riemann, and Mathias Berger

Thyrotropin (TSH), thyroxin (7"4), triiodothyronine (7"3),free T3 07'3), cortisol, prolactin, and human growth hormone (HGH) were measured every 2 hr during a night of sleep, the following day, and a night of sleep deprivation (5D) in 14 patient.; with major depressive disorder. In subgroups fl'4 (n = 5), reverse 7"3 (rT3), and luteinizing hormone (LH) (n = 6) were also investigated. Significant increases in TSH, T4, fl'4, T3, fT3, rT3, and cortisol and decreases in prolactin levels occurred during the night of SD, compared to the pattern during the night of sleep. The pre-SD 7"4 and T3 levels of the responders to SD were already higher than in the nonresponders, and increased less during SD. The cortisol and HGH concentrations of the responders rose higher during SD than those of the nonresponders. Changes in TSH and prolactin were not correlated to clinical response. Analysis of possible neurochemical mechanisms underlying this "pattern" of changes in different endocrine profiles suggests that enhanced noradrenergic activity might play a role in the changes in TSH, cortisol, thyroid hormones, and possibly HGH secretion during SD, and increased dopaminetgic tm,aeprobably induced the decline in prolactin levels. Additional effects of the serotonergic ~stem carnot be excluded at present. In conclusion, the data suggest that enhanced noradrenergic acsivity of the locus coeruleus stimulates alpha andlor beta adrenergic receptors in depressed patients during SD. This mechanism could we!! be involved in the antidepressant effect of this therapy.

Introduction In Part I of this series we reported increases in thyrotropin (TSH) and thyroid hormone concentrations at 8 AMfollowing sleep deprivation (SD) in depressed patients. The changes in TSH levels were correlated to clinical response. However, no significent changes in From the Psychiatrische Klinik und Poliklinik, Klinikum Rudolf-Virchow (Clmrlotlenburg) ~ Freien Universitit Berlin (A.B.); and the Max Pianck Institut ~ Psychiatric m Munich, FRG (M.B., D.R.). M.B. is now at the PsychiaUrische Klinik in Freibcrg and D.R. is at the Zentrelinstitut for Seeliscbe Gesundheit, Mmmbeim, FRG. .A_q~ ! y version of this paper was presented at the 42nd Annual Convemion of the Society of Biological Psychiatry, Chicago, May 6-10, 1987. Address reprint requests to Dr. A. BanmgeNner, Psychiatrische Klinik und Polildinik, Klinikum Rudolf-Virchow (Charlottenburg) der Freien Universittit Berlin, Eschenalle 3, D-1000 3eflin 19, Federal Republic of Gemumy. Received December 20. 1989; revised January 29, 1990. © 1990 Society of Biological Psychiat~j

~06-3223/90/$03.50

570

BIOL PSYC~HATRY 1990;28'569-587

A. Baumgartner et al.

prolactin and cortiso* concentrations were found at 8 AM after SD, which contradicts the reported rises in cortisol in responders (e.g., Gerner et al. 1979) and falls in prolactin during SD (Kasper et al. 1988). It is therefore not clear whether significant changes in other hormone axes did in fact orcur earlier during the night of SD which we were unable to detect as we drew our blood samples at 8 AM only. At about the same time as the study described in Part I was performed at the Psychiatrische Klinik in Berlin, another study with serial blood sampling over 36 hr during a night of sleep, the following day, and a night of SD was conducted at the Max Planck Institut fiir Psychiatrie in Munich. It therefore seemed expedient to analyze the results of this study together with the data presented in Part I. Methods Fourteen patients who met the Research Diagnostic Criteria (RDC) for major depressive disorder (definite) were included in the study. All patients were hospitalized on an open ward with a sleep research laboratory at the Max Planck Institut ffir Psychiatrie between May 1984 and June 1986. Eleven of the patients had endogenous subtype and 3 had nonendogenous subtype according to the RDC criteria. The patients were also diagnosed according to the DSM-III and ICD-9. According to the DSM-HI, 9 patients had major depression with me!Lncholia and 5 without mel~cbolia. According to the ICD-9, again 9 patients had endogenous depression and 5 neurotic depression. Or the 9 patients with endogenous depression, 5 had unipolar and 4 had bipolar subtype. Nine patients were women, 5 were men, and their mean age was 39.7 -+ 12.5 years (range 18-55). The criteria for inclusion in the study were the same as those given for the patient sample described in Part I. All the patients took part in a 19-day research projec:. All we:e drag-free for at least 7 days before the start of the study. Sleep was recorded for all patients between 11:00 PM and 7 AM for 14 consecutive nights [electroencephalogram (EEG) C3 - A2, C4 A I; horizontal electrooculograph (EOG), submental electromyog~.~un(EMG)]. The sleep stages were analyzed visually according to the criteria of Rechtschaffen and Kales (1968). After a night of adaptation to the ~leep laboratory, the cholinergic rapid eye movement (REM) induction test with the muscarinic agonist RS 86 was performed on nights 3 and 6, with placebo administration on nights 1, 2, 4, and 5. The experimental design and the results of these investigations are reported elsewhere (Berger et al. 1989). During nights 7 and 8, dream reports were elicited by REMP awakenings. On the ninth night, no special investigations were performed apart from sleep EEG recordings. On the tenth night the patients went to bed at about 10:30 PM and an intravenous cathetber was placed in an antecubital vein. It was kept patent with heparinized physiological saline. The first blood sample was drawn at 11 PM. The lights were turned out between 11:06 and 11:32 PM for all patients. During this night, sleep recordings were again performed and all patients were awakened at 6:30 AM. The mean total sleep time (TST) for all patients was 336.8 + 78.3 (range 172-421) rain. Only one patient had a TST of less than 200 min (172 rain). Three patients had TSTs between 200 and 300 rain, and the other 10 had TSTs of over 300 min. The mean sleep efficiency of all patients was 78.2% _+ 17o7% (range 41%-97%). The full results of the sleep analysis and the relationship between sleep parameters and. hormone secretion will be presented elsewhere. During the night of sleep, blood samples were drawn at 2-hr intervals (1 AM, ~ AM, and 5 AM) via a tube passed through a hole in the wall from the room next door. After

Neuroendocrinology of Sleep Deprivation

BIOL PSYCHIATRY 1990;28:~69-587

571

the patients had awakened (6:30.AM) further blood samples were drawn at 2-hr intervals during the day and following a night of SD until 7 AM. No daytime naps were permitted for the duration of the study and up to 8 PM following SD. The SD procedure was performed as described in Part I. All patients were rated on a six-item version of the Hamilton Rating Scale for Depression (HRSD) at 10 AM before a night of SD and at 10 AM after a night of SD. These ~,ix items had been taken from the original 21-item HRSD as follows: items 1 (depressed mood), 2 (feelings of guilt), 7 (work and activities), 8 (retardation), 10 (anxiety psychic), and i3 (somatic s/mptoms general). Response was defiried as an improvement of 30% or more on the 6-item HRSD scale on the day after SD (for explanation see Part I). On the basis of these criteria we obtained a group of 7 red,ponders and a group of 7 nonresponders. In order to assess the relationships between absolute changes in hormone concentrations during SD and the night of sleep and clinical response to SD respectively, Pearson's coefficient of correlation was calculated between the changes in the mean hormone concentrations on the 2 nights (mean of five measurements between 11 PM and 7 PM), and the changes in the HRSD scores (AHRSD) respectively. Cortisol and prolactin levels were measured in the samples of 13 patients and HGH concentrations in those of 12 patients. These determinations were performed by radioimmunoassay (RIA) in the laboratory of the Max Planck Institut. Cortisol was measured by RIA as described elsewhere (Krieg et al. 1987); the interassay v:~.ability was ~.9% at an average concentration of I0. I p,/dl. Prolactin and growth hormone were also determined by RIA (Serono). The interassay va,~.abi!i~, for prolactin was 10.6% and ~.6% for human growt.h hormone (HGH), at concentrations of 5.8 ng/ml and 3.2 ng/ml, respecfi,-e!y. All the o~er hormones were measured by the hormone research laboratory of the Klinikum Charlottenburg, Berlin. TSH, thyroxin (T4), triiodothyroein_e (T3), and free T3 (fT3) were measured for ~ll ! ~ Fatients, reverse T3 (rT3) and luteinizing hormone (LH) levels for 6, and fT4 levels for 5 patients. The hormones were measured in different numbers of patients because we did not have enough serum to measure all hormoncs in all samples. Details of the procedures used for determining the hormones evaluated in Berlin are g.xen in Part I. However, this time TSH was measured by the IRMAclon method (Henning, Berlin). The sensitivity of this method is 0.03 mU/liter. The interassay coefficient of variation measured continuously m'- urn- w.tNotatuly with uu, uwu w n u u , samples was 9.5% at a TSH concentration of 0. ! mU/!iter and 4.2% at a TSH concentration of 3.5 mU/liter. Details of the methods of data analysis used are given in Part I. In addition, we performed analysis of variance (ANOVA) for repeated measurements in order to assess whether there were significant differences in hormone secretion between the 2 nights (sleep and SD) for the whole group and between responders and nonresponders respectively. A multivariate ANOVA (MANOVA) was calculated with the dimensions "night effect" (sleep versus SD) and "time effect" (11 PM, I AM, 3 AM, 5 AM, and 7 AM) to evaluate differences in hormone secretion between the 2 nights. A second MANOVA with the dimensions "time effect" and "group effect" (responders versus nonresponders) was calculated for each night. In a third MANOVA, the interaction of the dimensioe~ "time effect, .... group effect," and "night effect" were determined to~e~er~ A preliminary report on the results of corti~ol, prolactin, and growth hormone in 9 of the 14 patients has already been published (Burger et al. 1985).

572

BIOL PSYCHIATRY 1990;28:569-587

A. Baumgartner et al.

Results

As already mentioned above, 7 of the patients we.re classified as responders and 7 as nonresponders. The two groups did not differ significantly with respect to age (responders) 42.7 4. 14.2; nonresponders 36.8 - 12.9) or initial severity of illness according to the HRSD (responders 10.7 4. 3.0; nonresponders 11.2 _ 3.9). However, the HRSD scores of the responders fell significantly after SD ( - 6 . 5 ; p = 0.01, Wilcoxon), wbeneas those of the nonresponders remained virtually unchanged ( + 0.5; ns). Three of the 5 patients with nonendogenous depression were classified as responders and 2 as nonresponders. The mean score for all 14 patients on the 6-item HRSD was 11.0 + 3.6 before SD and 8.0 4. 6.3 after SD. This difference was not significant (t = 2.1, p > 0.05). Comparisons of all mean hormone levels during the night of sleep and the night of SD are shown in Table 1. Table IA shows that the patterns of secretion of TSH, T4, fT3, and cortisol and prolactin oil ~he night of SD differed significantly from those on the nigh* of sleep. The differences for T3 approach significance (p = 0.06). The differences in hormone secretion at the individual sampling times are listed in Figures 16. Figure I shows that TSH is the only hormone whose mean level was already significantly higher at 11 PM during SD than on the night before (no SD). This is puzzling, as none of the patients were asleep at 11 PM during the first night (see Methods). During the night of SD all five thyroid hormone concentrations rose significantly, but to different extents: though the increase in T4 and fT3 were significant at all measuring times after 1 ! PM, the changes in T3 levels were significant at 3 AM only. Figure 2, however, reveals that these increases during SD mainly reflect the fact that the falls in these hormones during the preceding night of sleep do not occur during SD. The same applies to rT3 and fT4, which were measured in only 6 and 5 patients, respectively (Figure 3). Prolactin was found to decrease during SD at all measuring times after 11 PM, whereas no significant changes were seen for HGH or LH at any measuring time. Table IB shows that the secretion patterns of I'4, T3, and HGH of the responders and nonresponders differed significantly during the night of sleep. This was due to a higher mean T4 level during the first measuring period (night of sleep) and also during the day for the responders (62.0 4. 16.2 ng/ml for the night and 75.0 4. 18.2 ng/ml for the day) than for the nonresponders (57.2 -4- 9.2 ng/ml and 67.1 4- 10.9 ng/ml, respectively). 7he T3 levels of the responders were also higher before SD (see Fig-ore 5). The relationships between changes in hormone levels during the night of SD and changes in mood after SD are shown in Tab'le I and Figures 4-6. Table 1B does not show any significant differences between interactions between the time course and mean hormone values for the responders and nonresponders on the 2 nights. However, the p value for cortisol (p = 0.06) indicates a possible difference between the effects of SD on this hormone in responders and nonresponders. Figure 4 reveals an earlier nocturnal increase in cortisol during SD in responders. The individual results were as follows: a significant difference between the cortisol levels of the responders and nonresponders occurred at 1 AM (9.9 4- 5.0 ~g/dl for the responders versus 3.4 4- 4.6 I~g/dl for the nonresponders, p = 0.03) and 3 AM (responders 8.5 _+ 3.8 ~g/dl versus 5.3 4. 4.5 lzg/dl for tt,e noraesponders, p - 0.03, see Figure 4) and the mean cortisol levels of me responders during the night of SD were also significantly higher than those of the nonresponders (10.4 4- 2.7 Izg/dl versus 6.3 +_

Neuroendocrinology of Sleep Deprivation

toOL PSYCHIATRY

1990;28:569-587

573

Table 1A. Comparisons (ANOVA) of the Interactions Between Time Course and Mean H o r m o n e Values on the T w o Different Nights for the W h o l e Sample (n = 14)

TSH T4 T3 fr3 Cortisol PRL HGH

F

p

4.66 3.33 2.35 9.5i 10.37 6.84 1.09

0.003 ° 0.01 ° 0.06 0.0009 ~ 0.00001 ° 0.002 ° 0.37

Table 1B. Comparisons (ANOVA) of the Interactions Between Time Course and Mean Hormone Values for the Responders and Nom'esponders on Night i (a), Night 2 (b), and Both Nights (c) F TSH

T4

T3

fT3

(a) (b) (c) (a) (b)

(c)

1.20

(a) (b) (c) (a)

3.32 1.62 2.45 0.14 0.89

(b) (c) Cortisol

PRL

HGH

0.95 2.13 0.03 2.98 0.72

(a) (b) (c) (a) (b) (c) (a) (b) (c)

p 0.44 0.09 0.61 0.02 ° 0.57 0.28 0.01 ° 0.18 0.13 0.96 0.47

0.22

0.64

1.06 1.37 3.62 6.25 0.46 1.07 3.51 0.6O 0.31

0.38 0.25 0.06 0.89 0.75 0.32 0.01 ° 0.86 0.66

aSignificant results.

5.3 tLg/dl, p = 0.04). Significant increases in cortisol values were found in responders only at 1 AM and 5 AM, and not at all in the nonresponders (Figure 4). Pearson's coefficients of correlation between ATSH and AHRSD w~ e not significant. The TSH concentrations did not increase at all in 4 patients, 3 of whom were responders. Of note is the fact that 2 of these 3 patients had the lowest TSH concentrations of the whole group (mean TSH levels during sleep 0.01 mU/liter and 0.41 mU/liter). The third responder had the highest TSH levels in the group (mean TSH concentration during sleep 5.8 mU/liter). However, all the thyroid hormone concentratio~ of these patients were within the normal range. The mean T4 levels, which were higher in the rcsl:~r~de~ c ~ - . !:' ~.c~ ¢our.d

574

BIOLPSYCHIATRY 1990;28:569-587

TSH (mUll)

A. Baumgm-tner et al.

PRL

(ng/ml)

2.7-

sleep

sleep deprivation

13

2.62.5-

12

2.42.3-

11

2.2 2.1-

TSH (n = 14) 10

2.01.9,,

1.6-

•~

1.7

9 8

PRL (n = 13)

1.6 1.5

j

1.3

5

1.1

5

pm

|

am

I

I

pm

'e

I

am

II

,..- p

Neuroendocrinological investigations during sleep deprivation in depression. II. Longitudinal measurement of thyrotropin, TH, cortisol, prolactin, GH, and LH during sleep and sleep deprivation.

Thyrotropin (TSH), thyroxin (T4), triiodothyronine (T3), free T3 (fT3), cortisol, prolactin, and human growth hormone (HGH) were measured every 2 hr d...
2MB Sizes 0 Downloads 0 Views