Serial dexamethasone suppression tests in psychiatric illness: Part II. A study in major depressive disorder

Psychiatr_v Research,

2.5

18, 25-43

Elsevier

Serial Dexamethasone Suppression Tests in Psychiatric Illness: Part II. A Study in Major Depressive Disorder Andreas Received 1985.

Baumgartner,

Klaus-Jhgen

May 20, 198.5; revised version

and Irene

Giif,

received

September

Kihen

10, 1985; accepted

September

27.

Abstract. Weekly dexamethasone suppression tests (DSTs) were performed in 35 patients with major depressive disorder until clinical response. At initial evaluation, 65.7Yo of the patients showed nonsuppression, and 85.7% showed nonsuppression at least once during the treatment period. Normalization of the DST results usually coincided with or occurred before clinical recovery, although isolated “peaks” of DST nonsuppression occurred in 45.7% of the patients, irrespective of the clinical course. The test was not useful as a predictor of clinical recovery or relapse. Moderately ill depressed patients had significantly higher nonsuppression rates than schizophrenic or manic patients with corresponding severity scores, indicating that different factors might be associated with nonsuppression in different diagnoses. However, many abnormal DST results could neither be related to the course of the depression nor to the severity of illness; thus other factors must also be responsible for DST nonsuppression. Key Words. Longitudinal

dexamethasone

suppression

tests, depression.

In many previous studies, the dexamethasone suppression test (DST) has been described as a useful biological marker for endogenous depression (e.g., Carroll, 1982). There is, however, increasing evidence that patients with diagnoses other than melancholia frequently have abnormal DST results (for references see Part I of this report). While this evidence has led to rising doubts about the diagnostic reliability of the DST, its clinical relevance as a predictor of therapy response and relapse in major depression has been emphasized in several studies (Greden et al., 1980, 1982, 1983; Holsboer et al., 1982, 1983; Targum, 1984). Holsboer et al. (1982) conducted weekly DSTs in depressed patients and found that the results usually reverted to normal some weeks before clinical recovery, thus making the DST a possible predictor of therapy response. In another study, Holsboer et al. (1983) suggested that a change in DST results from normal to abnormal might indicate clinical relapse in patients who had

An earlier version of this report was presented at the 137th Annual Meeting of the American Psychiatric Association, Los Angeles, May 5-11, 1984. Andreas Baumgartner, M.D., is at the Psychiatrische Klinik und Poliklinik, Klinikum Charlottenburg, Freie Universitat Berlin. Klaus-Jiirgen C&f, M.D.. is Associate Professor of Internal Medicine at the Medizinische Klinik und Poliklinik (Hormone Research Laboratory), Klinikum Charlottenburg. Irene K&ten is Laboratory Assistant at the Hormone Research Laboratory, Klinikum Charlottenburg, Freie Universitat Berlin, Federal Republic of Germany. (Reprint requests to Dr. A. Baumgartner, Psychiatrische Klinik und Poliklinik, Klinikum Charlottenburg der Freien Universit’at Berlin, Eschenallee 3, 1000 Berlin 19, FRG.) 0165-1781:86:$03.50

0 1986 Elsevier Science Publishers

B.V

26

previously been in remission. In two similar studies, Greden et al. (1982, 1983) also found that the normalization of DST results coincided with or preceded clinical recovery. The results of Bowie and Beaini (1985) however, raised doubts about the importance of the DST as a reliabile predictor of clinical recovery, as these authors described a number of patients with fluctuating DST results, irrespective of the clinical course during a 6 week DST monitoring period. It nevertheless appeared reasonable to suppose that if the results of Greden, Holsboer, and their associates could be reproduced in a larger population, the DST might prove useful as a routine laboratory screening measure for patients with major depression. The test could then assist clinicians in making decisions on drug management for both inpatients and outpatients. Thus, the purpose of our study was to examine the pattern of weekly DST results in patients with major depressive disorder in the period between admission to hospital and time of clinical response. Methods Patients. All patients who were admitted to our six nonresearch wards between October 1983 and March 1984, and who met the Research Diagnostic Criteria (RDC) (Spitzer et al., 1978) for major depressive disorder (probable or definite) were considered for inclusion in the study (n = 77). Additional criteria for admission to the study were: (1) a score 2 18 on the 21-item Hamilton Rating Scale for Depression (HRSD) (Hamilton, 1960); (2) no history of alcoholism or drug abuse; (3) no history of severe weight loss (e.g., > 20% of the ideal body weight); (4) no medication that might affect the DST (Carroll, 1982; Greden et al., 1983); (5) no serious somatic illness; and (6) suitable veins for repeated venipuncture. Twenty-seven out of 77 patients were excluded for the following reasons: HRSD score < I8 (n q 5); history of alcoholism or drug abuse (n = 6); concomitant medication, such as cortisol, thyroxine, and carbamazepine (n = 5); concomitant diseases (n q 4); and no suitable veins for venipuncture (n = 7). Seven other patients did not give informed consent, five left the hospital within 2 weeks of admission, and three were excluded for organizational reasons. Thus, 35 patients (22 females, 13 males) completed the study. Their mean age was 54.7 (range 27-75) years. Patients were also diagnosed according to DSM-III (American Psychiatric Association, 1980) and ICD-9 (World Health Organization, 1978). All diagnostic classifications were made by one psychiatrist (A.B.). Thirty-two patients had endogenous depression (according to RDC), and 30 met criteria for melancholia (according to DSM-III). The low incidence of patients “without melancholia” (according to DSM-III. patient nos. 6, 7, 26, 3 I, 33 in Table 1) in our population is due to the fact that most of the patients with such a diagnosis fulfilled one or more of the exclusion criteria. Twenty-seven patients had unipolar depression, eight had bipolar depression (patient nos. 1, 5, 15, 18,20, 23,32, 35 in Table I), and seven had major depression with psychotic features (patients nos. 1, 5, 12, 14, 17, 27, 30). An electroencephalogram, an electrocardiogram, a routine laboratory screening, and a physical examination were done for all patients in order to exclude the possibility of major physical illness. Thirty-one of the patients were diagnosed as having major depression alone. One patient had mild diabetes mellitus, mild congestive heart failure, and mild organic brain syndrome (no. 24 in Table 1). Another patient had suffered a stroke of the middle cerebral artery 6 months before admission (no. 17 in Table I), and two others had mild organic brain syndrome (patients nos. 29, 30 in Table 1). Drug Treatment. Eighteen of the patients had received no psychotropic medication before hospital admission, and 17 had received antidepressants. neuroleptics, or minor tranquilizers. In these 17 patients, medication was withdrawn on hospitalization, and the first DST was conducted 3-4 days after admission in all 35 patients. Most of the patients (n q 27) were treated with either clomipramine or maprotiline; the dosage ranged from 100 to 200 mg/ day. If patients failed to respond clinically to one of these drugs within 4 weeks, they were put onto the other.

27 The drugs were administered orally or intravenously, and serum levels of both drugs were measured weekly for at least the first 4 weeks. All patients met laboratory standards for sufficient pharmacological treatment (e.g., Della Corte et al., 1979; Hollister, 1982). If they did not respond to clomipramine and maprotiline, patients were given another antidepressant medication, such as lofepramine, amitriptyline, or tranylcypromine. Three patients were receiving electroconvulsive therapy (patient nos. 1, 14,27 in Table I), and four were put onto lithium. Ten patients with psychotic depression or bipolar disorder were also on neuroleptics (patient nos. 5, 12, 14, 17, 19, 24, 25, 27, 30, 32 in Table 1). Three patients were treated by psychotherapy and were given no medication (patient nos. 6, 26, 33 in Table 1). All patients underwent a thyrotropin-releasing hormone (TRH) test every second week on the day before the DST. More detailed information on demographic characteristics, diagnoses, the individual drug treatment, and the temporal relationship between drug treatment and the DST is available from the authors on request. Clinical Ratings and Measurements. All patients were rated on the 21-item HRSD on the day of each DST until they showed clinical response. Clinical response was defined as an HRSD score < 10 and no longer fulfilling the criteria for major depressive disorder. Response to a specific form of therapy was defined as fulfillment of the above criteria within the first 4 weeks of treatment. The ratings were carried out in the afternoon, all by the same rater (A.B.), who was unaware of DST results until the patients responded clinically. Details of the control population, DST procedure, and data analysis are given in Part I of this report. In all patients, dexamethasone serum levels and weight were measured weekly. The results are presented in Part III.

Results Between Nonsuppressors and Suppressors. The distribution of postdexamethasone cortisol values is shown in Fig. 1. The nonsuppression rates for

Comparison

Fig. 1. Postdexamethasone

patients following

cortisol levels at 4 p.m. in healthy controls, and in hospital admission and after recovery tiea,ttIy COntrOII Tn = 67

Cortlrol 24 lugldll _ 22 2018l61412lo*6-

119?058

28

patients following admission (65.7%) and healthy controls (4.5%) differed significantly (x2 = 45.4, df= 1,p < 0.01).However, the nonsuppression rates for patients after recovery (15.6%) no longer differed significantly from those seen in the healthy volunteers (~2 3.6, df= 1, p > 0.05). If we follow Carroll (1982) in using a more “conservative” cutoff point (e.g., 5 pg/dl), the nonsuppression rate of the patients following admission (54.3%) still differs significantly from that of healthy controls (0%; x2 = 44.7, df= 1, p < 0.01).If we consider not only the initial DST, but all the results obtained over the entire study period, 85.7% of the patients showed nonsuppression at least once (68.8% at a cutoff point of 5 pg/dl). q

Pattern of Longitudinal DST Results. The relationships between the weekly DST results and the HRSD scores are shown in Table 1 and Fig. 2. We divided the patients into five different groups according to characteristic DST patterns. In group I (patient nos. I- 10 in Table 1-I), normalization of initially abnormal DST values coincided with (n = 3) or preceded (n 6) clinical recovery, depending on our cutoff point (2.35 pg/dl) and response criterion (HRSD score < 10). In one patient (no. 6) DST results returned to normal 1 week after clinical response. As described for the schizophrenic and manic patients, however, all these patients had already shown distinct clinical improvement in the week of DST normalization: the mean HRSD score of six patients q

Table l-l. Dexamethasone with recovery

suppression

test (DST)

responses

and HRSD

Patient no.

Weeks after initial DST

1

2

0

3.8

(32)

18.2

1

2.4

(18)

< 11

(12)

2

3.1

3

1.71 (10)

4

1.6

( 31

5

(28)

( 4)1
3

4

10.1 (32)

9.2

(23)

10.4

(23)

6 16.0

11.9 (191

6.1 117)

< 11 (11)
3.1 (171

3.1

6.7 (12)


IO)

< 11 (121

1.41


(0)

< 1

i 711


(01 IO1

4.0

/ 11)

5

< 11

( 411


01

6

(22)

13.5 (131

t 611 ( 81

7 r = 0.83

r = 0.95

r = 0.89

r = 0.79

DST week

df=3

df = 4

df=3

df=5

HRSD

NS

p = 0.002

p = 0.045

p = 0.033 r = 0.88

Pearson’s

r2

week

r = 0.89

r = 0.37

r = 0.96

DST week

df=2

df=3

df=2

df = 4

HRSD

NS

NS

p = 0.049

p = 0.016

Pearson’s

r3

week + 1

r = 0.91

r = 0.21

r = 0.38

r=l

DST week

df=

df = 2

df=

HRSD

NS

NS

p < 0.01

Pearson’s

r4

week + 2

1

1

df=3

p = 0.028

1. Plasma cortisol levels in &g/d1 and Hamilton Rating Scale for Depression (HRSD) scores in parentheses. Weeks of clinical recovery and of change to suppression/nonsuppression marked wrth a superscript 1. 2. Pearson’s r was calculated for DST results and HRSD scores of the same week IDST week - HRSD weeki. 3. Pearson’s r was calculated for DST results of 1 week and HRSD scores of the following week !DST weekHRSD week fl). 4 Pearson’s r was calculated for DST results of 1 week and HRSD scores 2 weeks later (DST week HRSD week +21.

29 whose DST normalized before clinical response had decreased by I 1.3 points (range 7-18). Thus, clinical improvement indicating possible therapy response was already evident and normalization of the DST had no relevance as regards continuation or alteration of drug treatment. Furthermore, it seems noteworthy that all 10 patients in this group were good therapy responders and no alteration in their treatment needed to be considered. As in Part I of this report, we again calculated Pearson’s r coefficient of correlation between the DST results and the HRSD scores of each patient to obtain further clarification of the temporal relationship between the two variables. A correlation was found for 8 of the 10 patients, and for four of them (nos. 4, 8, 9, and lo), the best correlation was found between the DST result of one week and the HRSD score of one of the two following weeks. It should be noted, however that in three of these four patients, there was also a good correlation between the DST result and the HRSD score of the same week, although it was less marked than the correlation mentioned above. The patients in group 11 (patient nos. 1 l-19 in Table I-II) also showed nonsuppression at the initial DST, but their pattern of DST nonsuppression could not be clearly related to clinical course. The only clinical relapse in these patients (patient no. 14 between weeks 3 and 4) was not predicted by the DST result. In only one patient scores.1 I. Normalization

of the DST 1-2 weeks before or coincident

Patient no. 7

8

10

Mean k SD cortisol

14.5 (221

11.0 i-4.9

27.0?

5.4

10

7.0 (20)

7.0 k 7.2

17.2 2

7.8

10

9

Mean f SD HRSD score

n

5.7 1221

6.2 (37)

15.8 (29)

3.3 I191

2.4 (26)

23.6

1.61 (14)

cl1

(251

10.1 134)

7.5 (21)

3.8 k 3.1

14.9 f 10.3

10

1.2 (15)


(111

15.1

3.5 (19)

3.6 t 4.6

12.0 +

6.2

9

1.2 ( 911

Cl

I 511

2.4 (171

1.31 (15)

1.3 !I0.5

7.52

5.9

8

1.8 I 61


(0)

2.21 (11)

(1

( 51'

1.3 -t0.5

3.62

4.5

6

(1

IO1

(1

i 5)1


IO)


(0,

(31) (21)

r= 0.76

r=

df=4

df=

df=6

df=5

df = 4

NS

p = 0.042

~=0.014

NS

D = 0.003

0.80

0.72

r=

0.91

r = 0.92

r=

df=3

df=4

df=5

df=4

df=3

NS

NS

o=O.OOl

NS

D = 0.026

r=

0.83

r=

5

r=

r = 0.97

r = 0.77

0.78

0.78

r = 0.97

r = 0.72

r= 0.88

r = 0.93

df=2

df=3

df = 4

df = 3

df=2

NS

p=O.O03

NS

p = 0.049

NS

r=

0.81

30 Fig. 2. Mean (k SD) cortisol concentrations (bars) and corresponding mean Hamilton Rating Scale for Depression (HRSD) scores (connected dots) for the total sample Pldsma Cortlsol (w/d11

HRSD SCOW

I\

12-l

35

30

l

10

25

!

8

.20

\ . 1 : lki .15

6

\!

4 I

24

10

-5

1

J-.-L_ -0 aftet response il wk 1 wk 2 wk 3 wk4 (n=351 (n=341 (n=34) (~1~331 (n=31l in=321 given for the week before treatment, the first 4 weeks of treatment, and the week after clinical

0

-

I

WK

Values

are

recovery.

(no. 11) did statistical analysis reveal a significant correlation between the DST result of one week and the HRSD score of the following two weeks. In four patients, the best correlation between the two variables appeared in the same week, and in four other patients, there was no such correlation at all-which raises doubts about the hypothesis that normalization of DST values generally precedes clinical recovery. It is noteworthy that in group II only one patient responded to the first drug administered and the mean time before clinical response for this group was 12.1 weeks, compared to 2.6 weeks for group I. Thus, these were the patients who were particularly difficult to treat and in whom a laboratory predictor of response or relapse could have been of much help to the clinician. However, it was precisely in these patients that the numerous “unexpected peaks” in the DST results excluded appropriate clinical decisions on the basis of the DST. The patients in group III (patient nos. 20-27 in Table I-111) all showed suppression at the first DST and changed to nonsuppression at least once during the study period. Statistical analysis did not reveal any significant correlation between DST results and HRSD scores in these patients. As in group II, the number of inexplicable “cortisol peaks” precludes clinical decisions on the basis of DST results. Group IV (patient nos. 27-30 in Table l-IV) comprised four patients whose DST values never reverted to normal. Three of them exhibited clinical recovery. Two of

31 these patients were also diagnosed as having mild organic brain syndrome, and their mean age was higher than that of all other patients (66.7 years and 53.0 years, respectively), although this difference was not significant (Mann-Whitney U 25, p 0.055). Group V (patient nos. 31-35 in Table 1-V) consisted of five patients who showed suppression at all testing times. This group was significantly younger than the total population (Mann-Whitney U = 26, p < 0.05). Patient no. 32 showed suppression during both the depressive and the manic phases of illness. In all 35 patients there was an obvious decrease in the mean postdexamethasone cortisol concentrations during treatment, and this decrease was associated with a similar progressive reduction in the mean HRSD scores (Table l-VI and Fig. 2). Screening of the individual correlations reveals that 7 out of 30 patients who exhibited nonsuppression on at least one DST showed the best correlation between the DST of one week and the HRSD score of one of the two following weeks. In seven other cases DST result and HRSD scorecorrelated best in the same week; in the other cases, there was no correlation at all. To determine whether the clinical usefulness of longitudinal DST evaluations could be improved by setting a higher cutoff point, we looked at the results obtained with a cutoff point of 5pg/dl. Six patients from groups I, II, and III then had to be classified as suppressors (group V). In six patients in group 1, the DST results became normal 1 week earlier, thus increasing the number of weeks by which the DST precedes clinical recovery, and in patient no. 7 there was a new “inexplicable peak” in the DST values. In group IV, one of the permanent nonsuppressors became a suppressor. Thus, the number of patients with confusing DST peaks was reduced from 16 to 11 by increasing the number of suppressors, and in group I the predictive value of the test was slightly improved. q

q

DST Results and Severity of Illness. At the initial DST, the mean HRSD score was 26.5 (SD 5.0) for nonsuppressors and 23.0 (SD 4.5) for suppressors. This difference was statistically significant (Mann-Whitney U 79, p < 0.05). The good correlation between the mean HRSD scores and mean cortisol levels of all patients during the study period and the good intraindividual correlations of cortisol concentrations and severity scores in some of the patients have already been described (Table 1, I-VI, and Fig. 2). However, interindividual comparison of DST results and the HRSD scores for the first week alone does not reveal any correlation (rz0.08, NS). If the interindividual and intraindividual data for all DSTs in the first 3 weeks are considered, the correlation becomes significant (r 0.35, p < 0.01). The result of the calculation for all the DSTs in all patients was also statistically significant (r = 0.35, p < 0.01) but cannot be considered absolutely correct, since different numbers of measurements were compared for each patient. We also calculated the percentages of nonsuppressors at different degrees of severity of illness. The patients were divided into four groups according to their HRSD scores: severely ill patients (HRSD score 3 31), distinctly ill patients (HRSD score 21-30), moderately ill patients (HRSD score 1 l-20), and clinical responders (HRSD score O-10). The calculation of the nonsuppression rates in the different groups again took into account all DSTs in all patients. A x2 analysis showed significant differences q

q

32 Table

l-11. Changes

in DST

response

independent

11

12

13

14

15

0

8.9 123)

8.0 (361

2.4 (281

9.1 (30)

23

(211

1

5.3 (22)

< 11 (28)

1.61 (26)

6.4 (26)

20.4

(171

2

1.71 (20)


(21)

1.2 (25)

3.2 (23)

3

2.41 (21)

~1

(23)

1.2 (161

1.81 (16)

21.6

(20)

4

2.21 (12)

5.81 (25)

1

9.71 (3.8)

21.9

(12)

5

1.3 ill)

(11

114)

1.7 (121

-

(36)

6

1.5 (11)

~1

(161

1.4 11O)l

-

(38)

12.2 ill)

7

1.4 i 311

~1

(181

1.4 i 91

-

(37)

21.7

8

~1

( 811

10.5' (121

9

~1

(6)

10


1111

1.11 1121 1.2 (151

9.0 (19)

9.7 (11) (16)

2.7 (37)

3.5 11.5)

-

(351

8.4 116)

10.8 (37)

1.21 (13)

11

1.1 (14)

-

12

1.7 IlOll

1.11 (25)


1.1 117)

12.41 (11)

13 -

14

(301

1.4 112) ( 9)1

~1

(23)

3.9 (12)

16

< 1

(15)

cl1

17

4.71 (11)

1.0 (16)

11 (181

12.11 116)

15

1.4 I 2)

9.4 Ill)

18

114)

19


(22)

17.8 116)

20

2.41 119)

10.3 (15)

21

1.31 (17)

2.7 (15)

22


4.2

23

2.61 (12)

24

3.4 1141

25

5.6 (15)

26

1.71 (181

413)

27

1.3 (13)

28

~1

( 8)1

29

2

I 71

Pearson's r2

r=

DST

df=6

week week

r=

DST

df=5

week

HRSDweek+l

NS

Pearson's r4

r=

DST

week

HRSDweek+2 2., 3., 4. See

0.65

r=

0.67

df=9

NS

Pearson's r3

i.,

course

Patient no.

Weeks after initial DST

HRSD

of clinical

4 = 0.023 0.65

r=

r = -0.09 df=

0.38

df = 8

NS

10

df=

19

df=4

df = 7

p=O.O43

NS l-l

I)= 0.009

df=

p = 0.017 0.23

0.51

I)= 0.001 0.51

NS

r=

df=22

r=

r=

Table

0.64

61'

df = 21

-0.04

NS

footnotesl-4.

r=

r=

r = 0.29

0.82

11

i

1.11 ( 31

r=

0.32

df=9

df=

18

NS

NS

r=

0.34

df=21 NS r=

0.35

df = 20 NS

in

33

nonsurmressors Patient no. 19

Mean f SD cortisol

4.4 (30)

4.1 (261

8.5 + 6.0

27.0 f 4.9

1.0 120)'

411

(20)

5.9 + 6.1

21.7 k 4.3

1

16

17

18

9.8 (28)

7.1 121)

5.4 (15)

6.4 121)

Mean k SD HRSD score

Cl1

(

411

2.4 (22)

1.0 (20)

126)

2.4? 2.6

20.0 + 6.4

< 1

11411

< 11 I221

1.1 (191

(1

(19)

3.6 i 6.8

18.9 f 3.0

< 1

[lo)1

2.41 (19)


115)


(251

5.1 * 7.0

18.6 Yk9.3


3.3 (231

~1

116)

(1

1181

2.5 + 3.0

14.3? 4.6

Cl141

Cl'

31.81 (181


i 51’

2.7 k 4.2

9.8 + 4.1


1.2 (14)

~11

(18)

(1

12)

3.7 +I7.2

10.4 k 7.0

1.6 114)

11

(151

Surgery

4.81 1161

4.41 (22) -

Cl1

(15)

< 1

(10)'

2.11 (19)



(1

(101

-

(12)

0.36

(31

112)

(21)

Monitoring

4118)

stopped

r = 0.89

r = 0.49

r= 0.50

r=

df=6

df=

df=

df=6

df=6

p=O.OOZ

NS

NS

p = 0.016

r=

10

12

NS

f= -0.31

r=

0.27

df=5

df=7

df=

11

NS

NS

NS

0.38

r = 0.00

r=

0.32

df=4

df=

7

NS

NS

r=

0.17

r = 0.79

r= 0.72

df=5

df=5

NS

NS

r = 0.05

r=

df=

df=4

df=4

NS

NS

NS

11

0.45

r= 0.53

n


1.2

2

3

1.1 ill) 1.3

1..2..3.,4See footnotes l-4,Table1-I.

NS

0.5

HRSD week+ 2

r=

df=4

f4

DST week

Pearson’s

NS

-0.04

HRSD week+

f=

df = 5

r3

DST week

Pearson’s

NS

NS

1

r = -0.09 df=

r= -0.12

df=6

NS

df=

NS

NS

1

r = -0.56

r = -0.43

NS

NS

df=6

r = -0.65 df=2

df=2

r = -0.26

df=6

df=5

NS

NS

8.2k6.4

11.8k3.8

12.5 5 5.0

NS

2.2 i- 1.3

3.1 ?I 2.4

3.3 2 3.6

r = 0.16

NS r = -0.47

NS

HRSD week

-

(161

Discharged

6.7

115,

15.2 t 4.3

22.7 + 4.9

Mean f SD HRSD score

r = -0.50

( 3)

t 8)’

3.6

3.5 & 4.5

1.3 2 0.5

Mean + SD cortisol

df = 3

1.4

1.9

1.51 113)

(21)

4.31 (13)

Cl

26

r = 0.33

( 511

(17)

i12)

(26,

19.41 (delirious)

1.0

25

df=8

NS

r= 0.13 df = 3

r = -0.03

df=8

r = 0.06

df=6

DST week

-

1.1 112,

2.7

Pearson’s

(12)

(151

1.81 111)

5.3

t 711 ( 01

10.4

1121

6.81 (4)

1.3

2.41

2.0

12.3 (17)

1.9 123)

24

9

-

(32)

1.3 (201

2.0

23

Patient no.

of clinical course in suppressors

10

1.5 1111 < 1 (12)

i 61’ (5,

~1
7 8

3.01 114)

115)

116)

c 11 114)

cl

2.1

(1

< 1 (17)

< 1 1181

1.0 (201

22

(111

< 11 (11)

1

(121

118)

4.51 (121


-

115)

(19,

6

Hypomanic

Hypomanic 9.41 -

-

131


<1

21

5

r2

~1

1

4

1.2 (18)

0

(81

20

Weeks after initial DST

Table l-111. Changes in DST results independent

5

6

6

6

7

n

35

Table l-IV. Nonsuppressors clinical recovery Weeks after initial DST

whose DST results failed to normalize

Patient no. 27

28

29

30

Mean + SD cortisol

Mean f SD HRSD score

n

0

15.4 126)

8.8 118)

23.1 (22)

19.6 (31)

16.7 !E6.2

24.3?

5.6

4

1

17.0 1241

13.7 114)

5.4 114)

17.4 (30)

13.42 5.6

20.52

7.9

4

2

21.4 (28)

11.7 (14)

3.1 (111

13.9 (31)

12.5k7.5

21.0flO.O

4

3

19.1 114)

10.5 (11)

2.5 (15)

4.7 (231

9.52 7.4

15.82

5.1

4

4

11.1 1271

7.0 (14)

3.0 ( 911

4.0 (151

6.4+ 3.5

16.32

7.6

4

5

16.8 1261

10.5

(121

3.1

6.0

9.1 i 6.0

12.82

9.2

4

6

17.5 (24)

10.4

I 9il

10.32 6.0

10.52

9.3

4

7

6.0 (12)

2.7 I 91

8

9.9 (11 )

6.0

9

12.0 (201

10

9.8

(

8)l

11

10.9 (281

12

14.6 (15)

13

18.2 (351

14

14.4

15

(

5)1

7.9 (121

16

7.9

17

9.8 (30)

18

4.0 (20)

19

6.6 (18)

(

8)1

20

9.7 (151

21

13.1 (251

22

3.4 (12)

23

Monitoring

I ( 8.5 ( ( 1.81 ( 4.01 ( 2.6 ( 10.6 ( 5.9 ( 13.8 (

( 2.8 (

61 41

( 10.4 (

7)' 5)

14.4( 4)

8)

8.5

7) 7) 71 7) 7) 6) 6) 61 6)

stomed Pearson's r2 DSTweek HRSDweek Pearson's r3 DSTweek

r= 0.45

r=

0.36

df=21

df=

15

p=O.O31

NS

r=

r=

0.58

df=20

df=

14

NS

p=O.O17

0.03

r=

r= 0.46

r=

df=5

df=6

df=5

p=O.O36

NS

p = 0.050

0.78

0.75

r = 0.49

r=

df=4

df=5

df = 4

NS

p=o.o15

p = 0.046

0.84

r= 0.82

HRSDweek +1 Pearson's r4 DSTweek

r=

-0.09

df=

19

r=

0.61

df=

13

r= 0.37

r = 0.96

r = 0.98

df=3

df=4

df=3

NS

p = 0.001

p = 0.001

HRSDweek f2 l.,

despite

NS

2., 3., 4. See footnotes

p=O.O14 1-4,

Table

1-I.

36 Table 1-V. Suppressors Weeks after initial DST

at all measurements

Patient no. 33

34

< 1 (30)

1.3 (21)

~1 (25)



32

31

0

< 1 (22)

1


2


3

< 1 ( 0)

< 1 Manic

< 1 ( 0)

4

< 1 ( 0)

< 1 Manic

< 1

( 1)’

1301

( 5)1
Mean + SD cortisol

35

(22)

Mean +_SD HRSD score n

~1

191

23.4 IL 4.3

5

(1

161

110

14.82

5


10.4 i

1.1 + 0.1

ii.9

161

l&O

< 1 141

l?O

3.8 +

6.8

4

< 1

151

l&O

5.0 2

7.0

4


141

Cl

7

14) 151

a

101

1)

5

< 1 1131


2)

<1(20) < 1

( 5) < 1

< 1 Manic

6


( 011

9 10

f

lit

1 Oi

a.5

5

101
Pearson’s

5)’ r = 0.83, df = 3, NS

r

See footnote 1. Table l-l.

Table l-VI. Mean postdexamethasone plasma cortisol concentration and corresponding mean HRSD scores for all Patients Mean cortisol

Mean I-fRSD score

n

6.7

25.3 + 5.1

35

6.2 + 6.7

1 a.0 2 7.4

34

Week

(I.rg/dU

0

7.42

1 2

3.9 2 4.6

16.0 + a.5

34

3

3.7 IL 5.2

13.3 * 6.6

33

4

3.3 * 4.3

11 .a f a.9

31

r2

Pearson’s

r = 0.94

DST week

df=3

HRSD

p=o.o13

week r3

Pearson’s

0.55

r = 0.98

DST week

df=2

HRSD week + 1

p = 0.016

P

Pearson’s

Mean r of all patients

0.49

r = 0.94

DST week

df=

HRSD week + 2

NS

0.45

1

2., 3., 4. See footnotes 2-4, Table l-l

between the nonsuppression rates of each group and those of any of the others (Fig. 3). To sum up, our data show evidence of a highly significant correlation between severity of illness and DST nonsuppression when intraindividual or both intraindividual and interindividual data are taken into account, but fail to show such a good correlation when the interindividual data are considered alone. Demographic

and Clinical

higher on the HRSD

Features

Our male patients scored rate than the female patients,

and DST Results.

and had a higher nonsuppression

37 Fig. 3. Nonsuppression rates and severity of illness in patients with major depression, calculated from all measurements during entire study period (n = 308) x2= 3 9,p,. 005

% Nonsuppressors

I

I

100

92.3

x*=7.7.

80

p x 0.01

I

I 64.5

60

40

26.1 20

0

II

Hamilton

-

_I-

O-10

11-20

21-30

> 30

(n=88)

(n=145)

(n=62)

(n=13)

Rating

Scale for Depression

(HRSDI

Score

but the difference was not statistically significant (Table 2). Unipolar patients had higher cortisol values and a higher nonsuppression rate than bipolar patients, but these differences also did not reach statistical significance (Table 2). Patients with psychotic depression had a significantly higher nonsuppression rate than those without psychotic features, but the difference between mean cortisol values and the HRSD scores of the two subgroups was not significant (Table 2). Nineteen of the patients responded clinically within the first 4 weeks of treatment. Sixteen of them responded to antidepressant drugs, two to psychotherapy, and one to electroconvulsive therapy. There were no significant differences between the nonsuppression rates or the postdexamethasone cortisol values of the responders and the nonresponders (Table 2). Comparison Between Patients With Major Depressive Disorder and Patients With Schizophrenia and Mania. There was no significant difference between the nonsuppression rates of the depressed patients and those of the patients with schizophrenia and mania (Part I of this report; x2 1.0, df = 1, p > 0.25). In both groups we found good intraindividual correlations between the severity of illness and DST results, manifested by a progressive fall in both mean DST values and mean severity scores during the study period. In both depressed and control patients, we q

38 Table 2. Demographic

Mean cortisol

pg/dIl

p (Mann-Whitney

(? SD)

score1

suppression

All patients

Female

Male

7.7 t 6.7

6.7 t 6.3

9.3 2 7.7

U = 117.5, NS

U)

Mean + SD HRSD

p (Mann-Whitney

and clinical features and dexamethasone

25.3 -t 5.1

24.5 + 4.5

26.6 i 5.9

U= 123, NS

U)

Nonsuppressors

23

13

Suppressors

12

9

3

59.1

76.9

Nonsuppression

rate (%I

65.7

P !X2)

10

9

= 1.15

p > 0.25 NS 1. Results of first DST. 2. Therapy response was defined as having a Hamilton Rating Scale for Depression start of drug treatment, electroconvulsive therapy, or psychotherapy

score 5 10 four weeks after

found a considerable number of cases in which “peaks” of DST nonsuppression could not be explained by the clinical course. If there is any difference at all between the patients in Parts I and II, it is between the nonsuppression rates for the moderately and distinctly ill patients in the two groups (Fig. 4). While the moderately and distinctly ill schizophrenic and manic patients (CGI 4-6) show a nonsuppression rate of only 16.9%, the nonsuppression rate in the depressed patients with a corresponding severity of illness (HRSD score 1 l-30) is 49.7% (~2 = 20.2, df= 1, p < 0.01). In other words, a schizophrenic patient must be more severely ill to become a nonsuppressor than a depressed patient. This reflects the fact that in Part I (schizophrenic and manic patients) there were no differences among the rates of abnormal DST values in the moderately ill patients, the distinctly ill patients, and the clinical responders, whereas in the sample of depressed patients there were significant differences among these groups. These calculations, however, must be discussed with caution, because the comparison between two different severity scales (HRSD and CGI) is somewhat arbitrary and the calculations are based on different numbers of evaluations in each patient. Moreover, there were large interindividual differences between the nonsuppression rates of the moderately and distinctly ill depressed patients. In some of them, nonsuppression was the exception (e.g., patient nos. 12, 13, and 18 in Table 1-II). Others showed abnormal results on almost all DSTs (patient nos. 15 and 27 in Table l-11 and Table I-IV). Finally the higher mean age of our depressed population, as compared to that of the patients with schizophrenia and mania, could be an additional source of error. Nineteen of our 32 therapy responders had final postdexamethasone cortisol values < 1 pg/dl. Eight further patients had DST results d 2.35 pg/ dl after clinical recovery. Thus the DST results in the recovered patients do not differ from those in healthy controls (see above), which indicates that the method of measuring cortisol is consistent and reliable within our study, and gives further justification for the low cutoff point. In Parts I and II of the study, we performed 475 DSTs in patients and healthy controls. No side effects of any kind were noted.

39 test (DST) response Psychotic

Nonpsychotic

10.6 k 5.2

Unipolar

6.9 + 7.0

8.2 ? 6.6

U = 59. NS 28.4 + 5.3

Bipolar 5.8 + 7.6

L/=85.

24.5 k 4.8

25.3 f 4.9

U = 54.5, NS

Therapy responders2

Therapy nonresponders

8.9 k 7.2

NS

6.1 ?c6.1

U = 109.5, NS

25.6 ? 5.9

26.3 + 5.1

U = 103, NS

24.2 2 5.0

U= 110, NS

7

16

19

4

13

0

12

8

4

6

6

68.4

62.5

100

57.1

70.4

50.0

10

x2 = 4.56

x2 = 1.14

x2 = 0.13

p < 0.05

p > 0.25

p > 0.5

NS

NS

Fig. 4. Nonsuppression

and severity of illness in schizophrenic bars) and in depressed patients (white bars)

rates

manic patients (hatched

and

% Nonsuppressors loogo80-

x2 = 2.4 n 5.

r

7060504030-

CGI

7- 81

HRSD The bars represent clinical responders ilefti. moderately and distinctly ill patients imiddle), and severely ill patients j right I. All dexamethasone suppression tests t DSTSI in all patients were included in the calculation.

40 Discussion Clinical Use of Weekly DSTs in Depression. We were unable to reproduce the findings of Holsboer et al. (1982, 1983), who used the DST as a predictor of therapy response or relapse in major depression. DST normalization preceded clinical recovery in only a minority of patients, and these were good responders whose condition had already improved at the time of DST normalization. In the patients who were more difficult to treat, the DST did not prove to be a predictor of clinical course. Our results are more consistent with those of Bowie and Beaini (1985), who described changes in DST results irrespective of the clinical course in 15 out of 36 depressed patients during a longitudinal DST study, and who also did not find the DST to be a predictor of relapse. As in our study, Greden et al. (1983) described patients with suppression at initial evaluation who later became nonsuppressors. Greden’s group also found DST peaks which were independent of clinical course, and described a patient whose DST did not return to normal despite clinical recovery. Although our results are similar to those of Greden et al., we cannot agree with the clinical advice of this group. The occurrence of the cortisol peaks without relation to the clinical course in a considerable number of patients, in particular, would seem to militate against the possibility of making clinical decisions on the basis of DST results. We were also unable to confirm the findings that DST results are predictive of good therapy response (Brown et al., 1979; Nelson et al., 1982; Ames et al., 1984). It is worthy of note that our results were obtained from patients accepted for the study in the order of hospital admission over a 6-month period and not from patients specially selected by a research unit. This underlines the lack of clinical usefulness of weekly DSTs under routine clinical conditions. Is a Nonsuppressor? In some previous studies, attempts have been made to validate nosological classifications by DST results (Coryell et al., 1982; Yerevanian et al., 1984). Other authors have found the DST to be a predictor of response to drug treatment in general or to a special kind of medication in particular (Brown et al., 1979, 1980; Nelson et al., 1982; Ames et al., 1984). In our study the considerable number of switches to DST nonsuppression in patients who originally showed normal DST results makes it hard to define whether these patients should be regarded as suppressors or nonsuppressors. Three studies meanwhile reported that DST results can change from suppression to nonsuppression from one episode to the next in the same patient (Brown and Qualls, 1982; Coryell and Schlesser, 1983; Grunhaus et al., 1983). Sherman et al. (1984) found that a patient can show both suppression and nonsuppression in the course of the same afternoon. For all these reasons, correlations and “nonsuppression” evaluated on the basis of between the variables “suppression,” only one initial DST result, and any other clinical variable, such as diagnosis or therapy response, should be regarded with caution.

Who

Depression, Nonspecific Stress, and DST Result. In Part I of this report we suggested that severity of illness or possibly nonspecific stress factors related to severity of illness may be important causes of DST nonsuppression in schizophrenic and manic patients. It can be seen from our longitudinal DST results for depressed

41 patients that the close intraindividual correlations between DST results and severity of illness found in a considerable number of patients could point toward severity of illness per se, stress, or a neurotransmitter disturbance specific for depression as causes of DST nonsuppression. We were unable to confirm reports that DST results regularly return to normal before the onset of clinical improvement (Holsboer et al., 1982) which would have been a strong argument against the stress hypothesis. The inexplicable cortisol peaks in 16 of our patients (Tables I-II and I-III), however, raised doubts about the DST result as a reflection of the underlying pathological neurochemistry of depression alone, since it is hard to explain why this “underlying disturbance” should change to normal or abnormal several times in the course of the episode in the absence of corresponding changes in the clinical course. In the case of our patients whose initially normal DST results became abnormal during the course of treatment (Table I-III), it is also difficult to understand why, if the DST results reflect an underlying pathophysiological process, they are normal at the beginning of the episode and become abnormal in the course of clinical improvement. With respect to the influence of severity of illness on DST results, we were able to show only a significant intraindividual correlation in over half of the patients, but no interindividual correlation between the two variables. The corresponding findings in the literature are contradictory (e.g.. Brownet al., 1979; Mendlewiczet al., 1982; Feinberg and Carroll, 1984; Bowie and Beaini, 1985). However, the “sensitivity” of the HPA axis could differ between patients: In one patient, an HRSD score of 20 could be enough to cause hypercortisolism, while another might have to be more severely ill to show the same effect. We therefore believe that even if there are no correlations between severity of illness and DST results, this does not rule out the possibility that severity of illness may play a role in DST nonsuppression in the individual case. Secondly, the conflicting results might be due to the inadequacies of the HRSD as regards measurement of overall severity of illness. Some items on this scale are overrepresented (e.g., hypochondriacal features), while others are neglected (depressive mood or depressive cognitions) (Sotsky, 1984). We therefore suggest (1) that calculation of the correlation between severity of illness and DST results should also include intraindividual measurements and (2) that different rating scales should be used to permit a more exact evaluation of severity of illness. It should be mentioned, however, that severity of illness could not have been responsible for all abnormal DST results, as we found “peaks” of cortisol nonsuppression that were independent of H RSD scores in almost half of the patients. Finally, one argument in favor of a causal connection between DST nonsuppression and depression has emerged from the study. A schizophrenic or manic patient must be more severely ill than a depressed patient to be a nonsuppressor. The qualities of stress which cause HPA axis overactivity are, however, far from being understood. Therefore, the possibility that those qualities are more closely related to depressive than to manic or schizophrenic symptoms cannot be excluded. This would mean that indirect stress effects are also implicated in the higher incidence of nonsuppression in depression. Apart from this, the differences between the nonsuppression rates for moderately and distinctly ill depressed patients and control patients must be regarded with caution for the reasons discussed above and still require further confirmation before any final conclusions can be drawn. The same applies to our finding that DST

42 nonsuppression occurs significantly more often in psychotic than in nonpsychotic depressed patients, as we cannot exclude the possibility that severity of illness is also at least partly responsible for this result. We conclude that severity of illness, irrespective of diagnosis, nonspecific stress, or both, is a possible factor that contributes to abnormal DST results in major depressive disorder. Furthermore, the data indicate that depression itself might be an additional factor in the origin of DST nonsuppression in some patients. A considerable number of abnormal DST results, however, could not be explained either by stress due to severity of illness or by depressive symptomatology. Therefore, the importance of the effects of other intervening variables on the DST (e.g., weight loss, dexamethasone serum levels, and age) is discussed in Part I11 of this report. References Psychiatric Association. DSM-III: Diagnostic 3rd ed. APA, Washington, DC (1980).

American Disorders.

Ames,

D., Burrows,

dexamethasone Psychiatry,

G., Davies, B., Maguire, test in hospitalized

suppression 144, 3 I I ( 1984).

and Sratistical

K., and depressed

Norman, patients.

Manual

of Mental

T. A study British

of the

Journal

Bowie, P.C.W., and Beaini, A.Y. Normalization of the dexamethasone suppression correlate of clinical improvement in primary depressives. British JournalofPsychialry,

qf

test: A 147,30

(1985).

Brown, W.A.. and Quails. Psychiarry

episodes.

Research,

B.C. Pituitary

adrenal

regulation

over multiple

depressive

7, 265 (1982).

Brown, W.A.,

Haier, R.J., and Qualls, C.B. Dexamethasone suppression test identifies of depression which respond to different antidepressants. Lancet, I, 928 (1980). Brown, W.A., Johnston, R., and Mayfield, D. The 24-hour dexamethasone suppression test in a clinical setting: Relationship to diagnosis, symptoms and response. American Journal of

subtypes

Psychiatry,

Carroll,

136, 543 (1979).

B.J. The dexamethasone

Psychiatry.

suppression

test for melancholia.

British

Journal

qf

140, 292 (1982).

Coryell, W., Gaffney, G.. and Burkhardt, B.E. DSM-III melancholia and the primarysecondary distinction: A comparison of concurrent validity by means of the dexamethasone suppression test. American Journal qf Psj,chiatr,,, 139, 120 (1982). Coryell, W., and Schlesser, M.A. Dexamethasone suppression test response in major depression: Stability across hospitalizations. Ps_vchiatry Research. 8, 179 (1983). Della Corte, L.. Broadhurst. A.D.. Sgaraghi, G.P., Filippini. S., Heeley, A.F.. James, H. D., Faravelli, C.. and Pazzagli, A. Clinical response and tricyclic plasma levels during treatment with clomipramine. British Journal qf Ps.t,chiatrj’, 134, 390 (1979). Feinberg. M.. and Carroll, B.J. Biological“markers” for endogenous depression. Archives qf General

PsychiatrJs,

41, 1080 (1984).

Greden, J.F., Albala, A.A.. Haskett, R.F.. James, N.Mcl., Goodman, L., Steiner, M., and Carroll, B.J. Normalization of dexamethasone suppression test: A laboratory index of recovery from endogenous depression. Biological Psj,chiatr_v, 15,449 (1980). Greden, J.F., de Vigne, J.P., Albala. A.A.. Tarika. J., Buttenheim. M., Eiser, A., and Carroll, B.J. Serial dexamethasone suppression tests among rapidly cycling bipolar patients. Biological Psychiatry,

17, 455 (1982).

Greden, J.F., Gardner, R., King, D., Grunhaus, L.. Carroll, B.J., and Dexamethasone suppression tests in antidepressant treatment of melancholia. Genera/

Psychiatr.v.

40,493

( 1983).

Kronfol,

Z.

Archives

of

43 Grunhaus, L., Greden, J.F., Carroll, B.J., and Shein, H. Thedexamethasone suppression in repeated hospitalizations. Biological Psychiatric, 18, 1497 (1983). Hamilton, M. A rating scale for depression. Journal of Neurology, Neurosurger.v. Psychiatry,

test and

23, 56 (1960).

L.E. Plasma concentration of tricyclic antidepressants in clinical practice. Journal 43, 66 (1982). Holsboer, F., Liebl, R., and Hofschuster, E. Repeated dexamethasone suppression tests during depressive illness. Journal of Affective Disorders, 4, 93 (1982). Holsboer, F., Steiger, A., and Maier, W. Four cases of reversion to abnormal dexamethasone suppression test response as indicator of clinical relapse: A preliminary report. Biological Hollister,

of Clinical

Ps.vchiauy,

Psychiatry,

18, 9

I I (1983).

Mendlewicz,. J., Charles, G., and Franckson, J.M. The dexamethasone suppression test in affective disorder: Relationship to clinical and genetic subgroups, British Journalqf’Ps~chiatr~,, 141,464

(1982).

Nelson, W.H., Orr, W.W., Stevenson, J.M., and Shane, S.R. Hypothalamic-pituitaryadrenal axis activity and tricyclic response in major depression. Archives of General P.s_vchialr.v, 39, 1033 (1982).

Sherman, B., Pfohl, B., and Winokur, G. Circadian analysis of plasma cortisol levels before and after dexamethasone administration in depressed patients. Archives c$General Psvchiatqs, 41, 271 (1984). Sotsky. S.M. The Hamilton Depression Rating Scale: A critique. Presented at the 137th Annual Meeting of the American Psychiatric Association. Los Angeles, May 5-I I (1984). Spitzer, R.L.. Endicott, J., and Robins, E. Research Diagnostic Criteria: Rationale and reliability. Archives of General Psychiarr.v, 35, 773 (1978). Targum, S.D. Persistent neuroendocrine dysregulation in major depressive disorder: A marker for early relapse. Biological P.sychiafry, 19, 30.5 (1984). World Health Organization. Mental Disorders: Glossary and Guide to Their Classification in Accordance With the 9th Revision qf’rhe International Classificarion of Diseases. WHO Geneva (1978). Yerevanian, B.I.. Privitera, M.R.. Milanese, E.. Sagi, E.A., and Russotto, J.C. The dexamethasone suppression test during recurrent major depressive episodes. Biological Psychiatry,

19, 404

(I 984).