Biological markers as classifiers for depression: A multivariate study

Pi-q.

N-Psychophammml

a wd

Psychlcrt Copyrt&t

kl-jplllOll

F’rhtcd

In Great

1994. 0

Vol.

18. pp. 898-9

1994 Elsevier Bdtah. 0278

Sclencc

All rt&s - 5546/S4

resewed $28.00

0276-5646(94)00064-6 BIOLOGICAL MARKERS AS CLA!3SlFIERS FOR DEPRESSION: A MULTIVARIATE STUDY

LlJC STANER 1,2, PAUL LINKOWSKI 1 and JULIEN MENDLEWICZ 1 1 Dept of Psychiatry, Erasme Hospital, Free University of Brussels, Brussels, Belgium (Final form,June 1993) Abstract Staner Luc, Linkowski Paul and Mendlewicz Julien: Biological markers as classifiers for depression: A multivariate study. Prog. NeuroPsychopharmacol. & Biol. Psychiat. 1994, 18(5): 899-914. 1. Delta TSH, REM latency, 4 pm and 11 pm post-dexamethasone cortisol values were determined after a wash-out period in a group of 74 non-selected depressed patients who were diagnosed (according to RDC with the SAX) as follows: 46 definite and 10 probable MD, 4 minor and 14 intermittent depression. 2. These biological variables, as well as gender, age and basal TSH were introduced in a principal component analysis. The four first PC scores explaining up to 77% of the data set were further calculated for each patients and used in a cluster analysis. A three clusters solution was retained. 3. DST escape and increased TSH response to TRH each identified subgroups of depressed patients. Conversely, blunted TSH response or REM latency were inefficient to classify patients. 4. Thus, HPA hyperactivity characterized CL-I patients (n=29). These were more severely depressed, displayed more endogenous features and were reported as being more anxious. 5. Increased TSH response to TRH identified CL-III, exclusively composed of female patients (n=lO) that displayed more apparent sadness and tended to be older. 6. In CL-II, the usual sex-ratio for depressive illness was reversed and patients (n=35) exhibited the least HPA axis disturbances and the same rate of blunted TSH response than in CL-I. They were also less severely depressed, displayed less endogenous characteristics and were rated as more mood reactive. 7 These results suggest heterogeneity in biological disturbances in depression and further stress the importance for controlling age, gender and severity of illnes in studies investigating biological markers in depression. 2Present address: Department of Psychiatry, Centre Hospitalier de Luxembourg, Luxemburg, Grand Duchy of Luxemburg

899

14 Ltd

900

L.Staner eta&

Kevwords: biological markers, depression, dexamethasone suppression test, REM latency, TRH test Abbreviations: Bipolar depression (BP), Cluster (CL), Dexamethasone Suppression Test (DST)Bpm (DST16) and llpm (DST23) post-dexamethasone cortisol; Hypothalamo Pituitary Adrenal (HPA), Hypothalamo Pituitary Thyroid (HPT), Major Depression (MD), Montgomery and Asberg Depression Scale (MADS); Newcastle Scale (NCS); Principal Component (PC), Rapid Eye Movement (REM), Research Diagnostic Criteria (RDC), Schedule for Affective Disorder and Schizophrenia (SADS), Thyroid Stimulating Hormone (TSH), Thyrotropin Releasing Hormone (TRH), Unipolar depression (UP). Introduction Considerable evidence suggest the presence of a biological substrate in depression; however, for almost 60 years, clinicians have failed to validly identify a biologically mediated depressive state, the so-called endogenous or melancholic subtype. Three main reasons have been advocated to explain this failure: 1) available scales supposed to delineate the endogenous profile are inaccurate; these scales do not consistently describe the same group of patients (Davidson et al., 1984) and, for a same schedule, non-uniform application of the diagnostic criteria across different research centers leads to significant differences in rates of endogenous depression (Zimmerman et al., 1990); moreover, internal consistency analyses do not support most definitions of endogenous depression (Young et al., 1986; George et al., 1989), 2) biological data used to validate these scales are inaccurate; these were progressively found to be of low specificity, to occur cross-diagnostically and could in fact relate to a particular nosogically aspecific psychological dysfunction (Van Praag et al., 1987). The related variables age, gender and global severity of illness might also confound studies investigating biological correlates of specific symptom profile (Loosen and Prange, 1982; Feinberg and Carroll, 1984), 3) the categorical definition between endogenous and non-endogenous depression is inaccurate; since Mapother (1926) and Lewis (1934) the qualitative (distinct illness) versus the quantitative

(illness

severity) distinction is a source of continuous and unresolved debate; however, to date, most frequently used definitions of endogenous depression embrace the qualitative approach; if untrue, this statement could have obscured the relationship between a biological variable and endogeneity as a dimension (Kendell, 1982).

BiologicalmarkersasclaasifWsfordepreasion

901

In view of these issues, in this study, the authors deliberately avoid the more usual procedure that consists in detecting biological disturbances in nosologically well defined patients. On the contrary, we try to delineate clinical characteristics that relate to groups of depressed patients sharing identical biological abnormalities. Thus, DST, TRH test and REM latency were used as classifiers of a group of 74 patients selected on basis of a sustained depressed mood whether or not they fulfilled major depression diagnostic criteria. The potential confounding effects of age, gender and illness severity, were taken into account in introducing it in the analyses as additional classifiers, Methods Subiects: The subjects were 74 patients (44 women and 30 men) aged 42 +/-14 years who underwent 3 consecutives recording nigths in our Sleep laboratory. Inform consent was obtained verbally. Case selection by applying diagnostic criteria was avoided and patients were included if, after a drug wash-out period of at least 10 days, they complained mainly about a depressed mood. Subjects were free of systemic illnesses or primary sleep disorders. Diagnoses were made according to RDC (Spitzer et al., 1977) with the SADS (Spitzer and Endicott, 1978) and were as follows: definite MD (n=46; 20 UP, 8 BP, 37 primary, 23 endogenous, 5 psychotic), probable MD (n=lO; 4 UP, 3 BP, 6 primary, 4 endogenous), minor (n=4; 2 UP and 1 BP) and intermittent depression (n=l4; 5 UP). Associated diagnoses were intermittent depression (n=9), alcoholism (n=8), drug use disorder (n=2) chronic minor depression (n=l), panic disorder (n=l) and schizotypal features (r&=1)for definite MD and panic disorder (n=4) for intermittent depression. There were no associated diagnoses for probable MD and minor depression. procedures: 8 variables were used to classify the patients group: age, gender, MADS scorest REM latency, DST16, DST23, basal and delta TSH. For each patient, the REM latency used in the analyses is the shortest of the values recorded during the 3 consecutives nights. Definitions of

L.Staner etal

902

sleep variables as well as procedure for sleep recording are described elsewhere (Kerkhofs et al., 1985). DST was performed according to Carrol et al. (1981) the day preceeding the first recording night and TRH test was completed following the overnight fast of the last recording night. Cortisol and TSH concentrations were measured by radioimmunoassay. According to Loosen and Prange (1982), delta TSH was calculated as the maximum increase above baseline in serum TSH, either 15 or 30 min after an intravenous injection of 0.5 mg of TRH. All diagnostic and psychometric assessments (MADS -Montgomery and Asberg, 1979-, NCS -Carney et al., 1965-, SADS and RDC diagnoses) were completed by the same investigator (SL) on admission to the sleep laboratory. Data Analysis: Logarithmic transformation were used for variables tested for departure from normality by means of the Kolmogoroff-Smirnov test. Relationships between variables were evaluated with Pearson's product-moment correlations. Independence of classification system was tested by means of contigency analyses (chi-square test with Yate's correction when appropriate). Between-group comparisons involving continuous data were computed using ANOVA; for MADS items, Kruskal-Wallis non-parametric ANOVA was applied. PC and cluster analyses were used as pattern recognition methods. A PC factor analysis with varimax rotation of the 8 classifying variables was first carried out to identify underlying uncorrelated dimensions in the variability of the biological data, controlling for the effects of age, gender and severity of illness. The resulting factors allowed the computation of PC scores that were calculated for each individual patient. These scores were then subjected to a cluster analysis (Ward's method) that generated disjunct clusters on the basis of unweighted classification variables. This procedure identified subgroups of patients on the basis of their PC scores. As these reflect uncorrelated dimension of their biological data, patients are gathered in subgroups sharing comparable biological characteristics.

Biological

903

markersasclassifierafordepresslon Results

Forminc Clusters Table 1 shows the correlation matrix of the five biological variables as well as age, gender and MADS scores. Age correlated with REM latency and delta TSH; gender with delta TSH; MADS with REM latency, delta TSH, DST16 and DST23; basal TSH with delta TSH; delta TSH with DST16 and DST16 with DST23. Table 1 Correlation Matrix of the Biological Variables, Age, Gender and MADS Scores Measured in 74 Depressed Patients Variable 1 R=

1 .oooo -0.0495 0.1834 -0.2465 tl -0.0019 -0.2204 * 0.0918 0.1324

key

2

1 .oooo 0.0745 -0.0148 0.1322 0.2470 * 0.0475 0.1041

4

3

5

0

7

8

I .oooo -0.3581 -0.0216 -0.2933 0.3480 0.2852

1.oooo

n *L

0.0987 0.1727 -0.0252 -0.0144

1 .oooo 0.5950 0.0403 0.0019

1 .oooo -0.2114 l -0.1773

1.0000 0.7816 -

1 .oooo

: variable 1 age; 2 gender (male=1 ; female=2); 3 MADS; 4 REM latency; 5 basal TSH; 6 detta TSH 7 DSTI6; 8 DST23 l-tailed significance: * .05 * .Ol -

,001

Results of the PC analysis are displayed in Table 2. The first four factors explaining 76.7% of the variance are represented. Factor 1 highly loaded on DST16 and DST23, factor 2 on basal and delta TSH, factor 3 on REM latency, MADS and age, and factor 4 on gender. Results of the cluster analysis with the corresponding PC scores calculated for each patient are illustrated in Fig 1 by the dendrogram. A three clusters solution was retained and, as indicated on Tables 3 and 4, they significantly differ on basis of, as expected, the classification variables gender, MADS score, DST16, DST23, basal and delta TSH.

9.19486 (.00585)

0.07432 (-.00201)

0.12262 (.f1718)

0.81699 (-57666) 0.64671 (38909) 0.62197 (.42332)

0.01689 (.OQ802) 0.23331 (-.04583)

0.86189 (60023)

0.09290 (-.13574) 0.17308 (.19544) 0.48860 (-.48793)

0.00549 (-.16202) 0.26055 (-.I 14)

0.00114 (-.03054) 0.02918 (-.00348)

11.7

0.93904

PC4

(1) Loading and (coefficient waigM). Note: the four PC scores ara computed as the sum of the several variables muitipliad by the appropriate coefficient weight of PCI, PC 2, PC3 and PC4, respectively.

Age

0.12166 (.01714) 0.15131 (-64867) 0.11436 (22903)

0.11070 (.17304) 0.37898 (.I 1OQ) 0.10356 (.00165)

REM latency MADS scorn

9.91776 (63033) a.81666 1.4762)

0.07965 (.07QQ) 0.20870 (-.OSQQS)

bTSH dmTSH

0.06117 (-.07928) 0.04752 (-.0863)

15.5

20.7

0.03112 (02428) 0.02697 (.02049)

1.23804

PC3

I.65997

0.93374 (.498X!) 0.92232 (.49287)

28.8

2.30112

PC2

DSTl0 DST23

Variablas (1)

Variance explained (%)

Eigenvalue

PC1

Results of the PC Analyis Performed with the Biological Variables, Age, Gender and MADS Scores Measured in 74 Depressed Patients. The First Four Varimax Rotated PCs ars Displayed

Table 2

Biological markers as clssslfiers for depression Table

905

3

Clinical and Biological Characterization of the Three Clusters Cluster (n=29)

Clinical and biological characteristics (1)

I

Gender, M/F 41.5 (13.8) 20.7 41.4 79.3 20105 4l21 13112 28.9 4.5 (2.3) 24.0 (10.1) 88.2 91.2 (80.9) 78.9 (38.5) 37.9 1.35 (0.58) 8.3 (3.4) 41.0 (21.8)

Age, Y Bipolar, % Unipolar recurrent, % RDC definite MDD, % Primary MDD, (yes/no) Psychotic MDD, (yes/no) Endogenous MDD, (yes/no) NCS Endogenous, % NCS score MADS score DST non suppression, % DSTIB, (pg/i) DST23, (~$0 Blunted TRH test, % bTSH, (mu/i) dmTSH, (mUIl) REM latency, (min)

Cluster II (n=35)

Cluster Ill (n=l0)

21114 39.5 (11 .Q) 14.3 37.1 45.7 18107 0135 13110 13.8 3.2 (2.1) 18.2 (8.9) 5.7 12.3 (13.2) 15.0 (14.4) 22.9 1.17 (0.51) 8.8 (2.8) 42.1 (28.1)

50.9 (21.3) 10.0 80.0 70.0 7101 l/08 3105 10.0 4.1 (1.5) 20.7 (8.4) 20.0 24.8 (29.4) 29.8 (35.1) 0.0 2.54 (0.80) 12.1 (3.8) 37.5 (18.5)

Chi-square or ANOVA significance co1 NS NS NS CO5 NS NS NS NS CO5 c.05 ~.OOOl ~.OOOl <.OOOl NS <.OOOl ~.OOOl NS

(1) mean and (standard deviation) are given for continuous variables

Table

4

MADS Items Scores in the Three Clusters

MAIN items (median and range) Apparent sadness Reported sadness Inner tension Reduced sleep Reduced appetite Concentration difficutties Lassitude Inability to feel Pessimistic thoughs Suicidal thoughts

Cluster I (n=29)

Cluster II (n=35)

2 3 3 3 1 2 3 2 2 1

2 2 2 3 0 3 3 1 1 0

(O-8) (O-8) (2-8) (O-8) (O-5) (O-8) (l-5) (O-S) (O-3) (O-4)

(O-4) (O-4) (l-4) (O-8) (O-4) (O-5) (0-S) (O-4) (O-3) (O-3)

Cluster Ill (n-10)

2.5 (O-4) 2 (O-3) 2 (O-5) 3.5 (2-5) 1.5 (O-4) 3 (O-4) 2.5 (2-3) 2 (l-3) 1 (O-8) 0.5 (O-3)

Kruskal-Wallis ANOVA stgnificance co1 c.05 co1 NS NS NS NS co5 NS NS

I

I

Fig 1 : Resultsof the clusteranalysis(Ward’s method) illustratedby fhe dendrogrem : the three major patientsclusters (I.11and Ill) are indicated. The clusteranalysiswas perfomed with the fhiit four PC scores calculated for each patient.

r

Biological markersas classifiers for depression

907

Cluster Characterization Other between-cluster notable differences were rate of definite MD, NCS score, and 4 MADS items scores (apparent or reported sadness, inner tension and inability to feel). Regarding biological variables, higher DST values characterized CL-I as higher delta TSH did for CLIII. These values were lower in patients beeing part of CL-II (n=35) and REM latency did not discriminate the clusters. Patients in CL-I (n=29) were more severely depressed and scored higher on NCS. Conversely, in CL-II that mostly consisted of male subjects, patients were less severely depressed, showed less endogenous features and had a lower rate of definite MD. CL-III was exclusively composed of female patients (n=lO) that tended to be older. Concerning MADS items, reported sadness and inner tension were highly scored in CL-I as was apparent sadness in CL-III. Inability to feel was less severely scored in CL-II. Discussion The present results confirm previous studies supporting the hypothesis of a biological heterogeneity in depression. To our knowledge, only

two studies investigated simultaneously DST, TRH test

and REM latency in the same group of depressed patients and found coexisting disturbances in less than 10% of their samples (Rush et al., 1983; Hubain et al., 1991). Our results are also on line with several reports that demonstrate no clear association between these three biological markers (Berger et al., 1982; Targum et al., 1982; Kupfer et al., 1984; Giles et al., 1987; Pugnetti et al., 1987; Barry and Dinan, 1990). The lack of interdependence of these biological disturbances in depression is further confirmed in our PC analysis. In fact, controlling for the effects of gender, age and severity of illness, REM latency, HPA and HPT axis linked values loaded each separately on different factors. Interestingly, our PC builded clusters did relate to only one of the usual biological marker of depression, i.e. DST non suppression in CL-I. REM latency was not discriminant and increased rather than blunted TSH response to TRH characterized one cluster (CL-III). Moreover, usual clinical subtypes of depressive illness, such as endogenous, psychotic, primary, BP or UP failed to differentiate the three clusters.

L.Staner et&

908

Severitv. Endoaeneitv. Anxietv and HPA axis Hyperactivity: HPA axis hyperactivity clearly distinguishs CL-I patients. These were more severely depressed (as assessed by definite MD rates or by total MADS scores or scores on item 2, reported sadness) and displayed more NCS endogenous characteristics. Moreover, they were reported as more anxious on MADS item 3 (inner tension). These findings are on line with studies linking depression severity (Maes et al., 1986; Whiteford et al., 1987), endogenous features (Feinberg and Carroll, 1984; Kumar et al., 1986; Winokur et al., 1987) and anxiety (Rubin et al., 1987; Meador-Woodruff et al., 1990) to HPA axis hyperactivity. Conversely, CL-II patients, who exhibited the least HPA axis disturbances, had also the lowest definite MD rates, the lowest MADS or NCS scores and were rated as more mood-reactive (MADS item 8: inability to feel). Ansseau et al. (1987) demonstrated a higher proportion of DST suppression in male minor but not major depressive patients. This is to be compared with the unusual sexratio observed in CL-II that contains, in more than half the cases, patients having minor forms of depression (i.e., probable MD, minor or intermittent depression). Severitv, Endoaeneitv and HPT Axis: Although a significant negative correlation was observed between MADS scores and delta TSH, a blunted response did not distinguish the more severely depressed patients. Such a negative correlation was also previously reported by studies that investigated both major and non-major depressive patients (Extein et al., 1981; Maes et al., 1989) but the prevalent view does not associate blunted response and depression severity, at least in major depression (Loosen and Prange, 1982; Linkowski et al., 1983; Vanelle et al., 1990). Unlike HPA hyperactivity, HPT disturbances failed to identify endogenously defined patients. In this regard, several studies linked abnormal TRH test results either to psychomotor agitation, violent suicidality and anxiety or panic attacks (Kirstein et al., 1982; Linkowski et al., 1983; Calloway et al., 1984; Gillette et al., 1989; Corrigan et al., 1992; Staner et al., 1992) rather than to the endogenous subtype.

Biologicalmarkersas classiflersfordepression

909

Increased TSH Resnonse, Gender and Aae. Increased TSH response to TRH, a picture likely to occur in 9 to 17% of depressed subjects (Winokur et al., 1983; Targum et al., 1983; Fava et al., 1992), characterizes CL-III patients. As these were exclusively females and tended to be older, this finding lends support to studies that report increased TSH responses preferentially in female (Winokur et al., 1983; Baumgartner et al., 1986) or in some aged depressed patients (Wahby et al., 1988; Targum et al., 1992). Indeed, according to several authors (Baumgartner et al., 1986; Maes et al., 1989; Vanelle et al., 1990), age and delta TSH were negatively correlated in our overall depressed sample. Our cluster analysis thus clearly identifies a subgroup of elderly depressed women that depart from this general propensity. Wahby et al. (1988) suggested that these anomalies could identify depressed patients with subclinical hypothyroidism that may have contributed to the depression. REM Latency: Compared to common definitions of normal values, REM latency was found equally decreased in the three clusters and could thus not separate them. In a recent review and meta-analysis of 177 studies with data from 3689 affectively ill patients, Benca et a1.(1992) concluded that REM latency weakly discriminates between various depressive subtypes (including the primary, endogenous and psychotic distinctions) except for three of them. Only bipolar, nonendogenous and secondary depressive patients had prolonged REM latency compared with primary or mixed groups of depressed patients. Our results with REM latency are also to be compared with other studies that were not controlled for age, gender or severity of illness. Shortening of REM latency seems to relate to psychotic symptoms in depression (Ansseau et al., 1984a; Kupfer et al., 1986; Thase et al., 1986). Some studies reported up to 50 % of concordance in DST non suppression and shortened REM latency in endogenous depressives (Ansseau et al., 1984b; Mendlewicz et al., 1984). Others (Asnis et al., 1983; Kerkhofs et al., 1986) also linked very short REM latencies (i.e., less than 20 min) with DST non suppression in this depressive subtype.

L.Staner etaf.

910

Conclusion To

summarize, this study supports the concept of a biological

heterogeneity in depressive illness. Our analyses suggest that HPA, HPT and REM latency abnormalities encountered in depression do not cluster around a unique group of patients considered to suffer from a unique biologically mediated depressive subtype often related to the endogenous or melancholic subtype. Instead, and taking into account the effects of major counfounding variables in biological research, our results suggest that DST escape and increased TSH response to TRH each identifies subgroups of depressed patients that could be further characterized on basis of illness severity, symptoms features and sex-ratio. Moreover, blunted TSH response or REM latency alone were inefficient to classify depressed patients in our sample. References ANSSEAU M, SCHEYVAERTS M, DUMONT A, POIRRIER R, LEGROS JJ and FRANCK G (1984a): Concurrent use of REM latency, dexamethasone suppression, clonidine and apomorphine tests as biological markers of endogenous depression: a pilot study. Psychiatry Res &?: 261272. ANSSEAU M, KUPFER DJ, REYNOLDS CF III and MC EACHRAN AB (1984b): REM latency distribution in major depression: clinical characteristics associated with sleep onset REM periods. Biol Psychiatry 19: 16511656. ANSSEAU M, DEPAUW Y, CHARLES G, CASTRO P, D'HAENEN H, DE VIGNE JP, HUBAIN P, LEGROS JJ, PELC I, TOSCANO A, WILMOTTE J and MENDLEWICZ J (1987): Age and gender effects on the diagnostic power of the DST. J Affective Disord 12: 185-191. ASNIS GM, HALBREICH DAVIS M, ENDICOTT secretion and REM J Psychiatry &Q:

U, SACHAR EJ, NATHAN RS, OSTROW LC, NOVACENKO H, J and PUIG ANTICH J (1983): Plasma cortisol period latency in adult endogenous depression. Am 750-753.

BARRY S and DINAN TG (1990): Neuroendocrine challenge tests in depression: a study of growth hormone, TRH and cortisol release. J Affective Disord 18: 229-234. BAUMGARTNER A, HAHNENKAMP L and MEINHOLD H (1986): Effects of age and diagnosis on thyrotropin response to thyrotropin-releasing hormone in psychiatric patients. Psychiatry Res u: 285-294.

Biological markeraasclassitlers for depression

911

BENCA RM, OBERMEYER WH, THISTED RA and GILLIN JC (1992): Sleep and psychiatric disorders. A meta-analysis. Arch Gen Psychiatry fi: 651-668. BERGER M, DOERR P, LUND R, BRONISH T and VON ZERSSEN D (1982): Neuroendocrinological and neurophysiological studies in major depressive disorders: are there biological markers for the endogenous subtype ? Biol Psychiatry u: 1217-1242. CALLOWAY SP, DOLAN RJ, FONAGY P, DE SOUZA VFA and WAKELING A (1984): Endocrine changes and clinical profiles in depression. II. The thyrotropin-releasing hormone test. Psycho1 Med 14: 759-765. CARNEY MWP, ROTH M and GARSIDE RF (1965): The diagnosis of depressive syndromes and the prediction of E.C.T. response. Br J Psychiatry 111: 659-674. CARROLL BJ, FEINBERG M, GREDEN JF, TARIKA J, ALBALA AA, HASKETT RF, JAMES NM, KRONFOL 2, LOHR N, STEINER M, DE VIGNE JP and YOUNG E (1981): A specific laboratory test for the diagnosis of melancholia: standardization, validation, and clinical utility. Arch Gen Psychiatry a: 15-22. CORRIGAN MHN, GILLETTE GM, QUADE D and GARBUTT JC (1992): Panic, suicide, and agitation: independent correlates of the TSH response to TRH in depression. Biol Psychiatry 31: 894-892. DAVIDSON JRT, TURNBULL CD, STRICKLAND R and BELYEA M (1984). Comparative diagnostic criteria for melancholia and endogenous depression. Arch Gen Psychiatry 41: 506-511. EXTEIN I, POTTASH ALC and GOLD MS (1981): The thyrotropin-releasing hormone test in the diagnosis of unipolar depression. Psychiatry Res 5: 311-316. FAVA M, ROSENBAUM JF, BIRNBAUM R, KELLY K, OTTO MW and MC LAUGHLIN R (1992): The thyrotropin response to thryrotropin-releasing hormone as a predictor of response to treatment in depressed outpatients. Acta Psychiatr Stand &: 42-45. FEINBERG M and CARROLL BJ (1984). Biological markers for endogenous depression: effect of age, severity of illness weight loss and polarity. Arch Gen Psychiatry 41: 1080-1085. GEORGE LK, BLAZER DG, WOODBURRY MA and MANTON KG (1989): Internal consistency of DSM-III diagnoses. In: The Validity of Psychiatric Diagnoses, LN Robins and JE Barrett teds), pp 99-123, Raven Press, New York. GILES DE, SCHLESSER MA, RUSH J, ORSULAK PJ, FULTON CF and ROFFWARG HP (1987): Polysomnographic findings and dexamethasone nonsuppression in unipolar depression: a replication and extension. Biol Psychiatry a: 872-882.

L.Staner et&.

912

GM, GARBUTT JC and QUADE DE (1989): TSH response to TRH in depression with and without panic attacks. Am J Psychiatry 146: 401-40s.

GILLETTE

HUBAIN P, STANER L, DRAMAIX M, RIELAERT C, PAPADIMITRIOU G, MENDLEWICZ J and LINKOWSKI P (1991): Neuroendocrine tests in major depression: relationship to clinical and sleep EEG variables. Biol Psychiatry 29: 3338-3348. KENDELL RE (1982): The choice of diagnostic criteria for biological research. Arch Gen Psychiatry 2: 1334-1339. KERKHOFS M, HOFFMAN G, DE MARTELAERE V, LINKOWSKI P and MENDLEWICZ J (1985): Sleep EEG recordings in depressive disorders. J Affect Disord 9: 47-53. KERKHOFS M, MISSA JN and MENDLEWICZ J (1986): Sleep electroencephalographic measures in primary major depressive disorder: distinction between DST suppressor and nonsuppressor patients. Biol Psychiatry 21: 228-232. KIRSTEIN L, GOLD MS, EXTEIN I, MARTIN D and POTTASH ALC (1982): Clinical correlates of TRH infusion test in primary depression. J Clin Psychiatry 43: 191-194. KUMAR A, ALCSER K, GRUNHAUS L and GREDEN JF (1986): Relationships of the dexamethasone suppression test to clinical severity and degree of melancholia. Biol Psychiatry 2: 436-444. KUPFER DJ, JARRETT DB and FRANK E (1984): Relationships among selected neuroendocrine and sleep measures in patients with recurrent depression. Biol Psychiatry 19: 1525-1536. KUPFER DJ, REYNOLDS CF III, GROCHOCINSKI VJ, ULRICH RF and MC EACHRAN AB (1986): Aspects of short REM latency in affective states: a revisit. Psychiatry Res 17: 49-59. LEWIS AJ (1934): Melancholia: a clinical survey of depressive states. J Ment Sci &Q: 277-378. LINKOWSKI P, VAN WETTERE JP, KERKHOFS M, BRAUMAN H and MENDLEWICZ J (1983): Thyrotrophin response to thyrostimulin in affectively ill women: relationship to suicidal behaviour. Br J Psychiatry 138: 401-404. LOOSEN PT and PRANGE AJ (1982). Serum thyrotropin response to thyrotropin-releasing hormone in psychiatric patients: a review. Am J Psychiatry 139: 405-416. MAES M, DE RUYTER M, HOBIN P and SUY E (1986): The dexamethasone suppression test, the Hamilton depression rating scale and the DSMIII depression categories. J Affective Disord lo: 207-214.

Biological markers as classtiers for depresslon

913

MALES M, VANDEWOUDE M, NAES L, SCHOTTE C and COSYNS P (19893: A revised interpretation of the TRH test results in female depressed patients. Part I: TSH responses. Effects of severitv of illness. thyroid hormones, monoamines, age, sex hormonal, co;ticosteroid'and nutritional state. J Affective Disord a: 203-213. MAPOTHER E (19261: Discussion on manic-depressive psychosis. Br Mad J 2: 872-876. MEADOR-WOODRUFF JH, GREDEN JF, GRUNHAUS L and HASKETT RF (1990): Severity of depression and hypothalamic-pituitary-adrenal axis dysregulation: identification of contributing factors. Acta Psychiatr Stand u: 364-371. ~NDLEW~CZ J, KERKHOFS M, HOFF~NN G and LINKOWSKI P (1984): Dexamethasone suppression test and REM sleep in patients with major depressive disorder. Br J Psychiatry 145: 383-388. MONTGOMERY SA and ASBERG M (1979): A new depression scale designed to be sensitive to change. Br J Psychiatry 134: 382-389. PUGNETTI L, SCARONE S and BRAMBILLA F (1987): Biological heterogeneity of major depressive disorder: indications by sleep EEG and TRH stimulation test findings. Neuropsychobiology &Z: 124-

RUBIN RT, POLAND RE, LESSER IM, MARTIN DJ, BLODGETT ALN and WINSTON RA (1987): Neuroendocrine aspects of primary endogenous depression III. Cortisol secretion in relation to diagnosis and symptom patterns. Psycho1 Med a: 609-619. RUSH AJ, SCHLESSER MA, ROFFWARG HP, GILES DE, ORSULAK PJ and FAIRCHILD C (1983): Relationships among the TRH, REM latency and dexamethasone suppression test: preliminary findings. J Clin Psychiatry & (set 2): 23-29. SPITZER RL, ENDICOTT J and ROBINS E (1977):. Research Diagnostic Criteria for a selected group of functional disorders, third edition, pp. New York State Psychiatric Institute, New York. SPITZER RL and ENDICOTT J (1978): A diagnostic interview. The schedule for affective disorders and schizophrenia. Arch Gen Psychiatry II: 837-844. STANER L, MAES M, BOUILLON E and LINKOWSKI P (1992): Biological correlates of the Newcastle scale in depressive illness: a multivariate approach. Acta Psychiatr Stand 85: 345-350. TARGUM SD, SULLIVAN AC and BYRNES SM (1982): Neuroendocrine interrelationships in major depressive disorder. Am J Psychiatry 139: 282-286.

L.Staner etaL

914

TARGUM SD, SULLIVAN AC and BYRNES SM (1983): Compensatory pituitarythyroid mechanisms in major depressive disorder. Psychiatry Res 6: 85-96. TARGUM SD, MARSHALL LE and FISCHMAN P (1992): Variability of TRH test response in depressed and normal elderly subjects. Biol Psychiatry 31: 787-793. THASE ME, KUPFER D3 and ULRICH RF (1986): Electroencephalographic sleep in psychotic depression: a valid subtype. Arch Gen Psychiatry 43:

886-893.

VANELLE J-M, POIRIER MF, BENKEFALT C, GALINOWSKI A, SCHECHTER D, SUZINI DE LUCA H and LOO H (1990): Diagnostic and therapeutic value of testing stimulation of thyroid-stimulating hormone by thyrotropin-releasing hormone in 100 depressed patients. Acta Psychiatr Stand a: 156-161. VAN PRAAG HM, KAHN RS, ASNIS GM, WELTZER S, BROWN SL, BLEICH A and KORN ML (1987): Denosologization of biological psychiatry or the specificity of 5-HT disturbances in psychiatric disorders. 3 Affective Disord u: 1-8. WAHBY VS, IBRAHIM GA, GILLER EL, MARTIN RP, SADDXR FW, SINGH SP and MASON JW (1988): Relationship of age to TSH response to TRH in depressed man. Acta Psychiatr Stand D: 283-288. WHITEFORD HA, PEABODY CA, CSERNANSKY JG, WARNER MD and BERGER PA 11987): Elevated baseline and postdexamethasone cortisol levels. A reflection of severity or endogeneity ? J Affective Disord g: 199202. WINOKUR A, AMSTERDAM YD, OLER J, MENDELS 3, SNYDER PJ, CAROFF SN and BRUNSWICK DJ (1983): Multiple hormonal responses to protirelin (TRH) in depressed patients. Arch Gen Psychiatry a: 525-531. WINOKUR G, BLACK DW and NASRALLAH A (1987): DST nonsuppressor status: relationship to specific aspects of the depressive syndrome. Biol Psychiatry 2: 360-362. YOUNG MA, SCHEFTNER WA, KLERMAN GL, ANDREASEN NC and HIRSCHFELD RMA (1986): The endogenous subtype of depression: a study of its internal construct validity. Br J Psychiatry 148: 257-267. ZIMMERMANN M, CORYELL WH and BLACK DW (1990): Variability in the application of contemporary diagnostic criteria: endogenous depression as an example. Am J Psychiatry 147: 1173-1179. Correspondence and reprint requests should be addressed to: Dr. Luc Staner Department of Psychiatry - Centre Hospitalier de Luxembourg Rue Barblg, 4 - L-1210 Luxemburg (Grand Duchy)