The predictive validity of atypical neurovegetative depressive symptoms identified by the first principal component in the DUAG trial of moclobemide versus clomipramine

Journal of Affective Disorders 140 (2012) 253–259

Contents lists available at SciVerse ScienceDirect

Journal of Affective Disorders

j ourna l homepage: www.e lsev ie r .com/ locate / jad

Research report

The predictive validity of atypical neurovegetative depressive symptomsidentified by the first principal component in the DUAG trial of moclobemideversus clomipramine

P. Bech a,⁎, K.B. Stage b, J.K. Larsen c, P. Vestergaard d, L.F. Gram e

and The DUAG 1

a Psychiatric Research Unit, Mental Health Centre North Zealand, Copenhagen University, Hillerød, Denmarkb Department of Psychiatry, Odense, University Hospital, Odense C, Denmarkc Department of Psychiatry, Mental Health Centre Ballerup, Department Gentofte, Denmarkd Department of Psychiatry, Aarhus University Hospital, Århus, Denmarke Dep. Clinical Pharmacology, IST, University of Southern Denmark, Odense, Denmark

a r t i c l e i n f o

⁎ Corresponding author at: Psychiatric Research Unifax: +45 48 26 38 77.

E-mail address: [email protected] (P. Bech).1 DUAG centre heads (1993): Kragh-Sorensen P (Dep

Hospital); Bolwig TG (Department of Psychiatry, RigshHospital); Larsen, JK (Department of Psychiatry, Frederi

0165-0327/$ – see front matter © 2012 Elsevier B.V. Adoi:10.1016/j.jad.2012.02.011

a b s t r a c t

Article history:Received 22 November 2011Received in revised form 6 February 2012Accepted 6 February 2012Available online 29 February 2012

Objective: To investigate to what extent the primary depression subtype atypical depressioncan predict differential outcome of the mono-amino-oxidase inhibitor (MAO-I) moclobemideand the tricyclic antidepressant clomipramine in the Danish University AntidepressantGroup Study (DUAG).Methods: In a randomised, double blind trial, a total of 117 patients with major depression weretreated over 6 weeks with either 400 mg moclobemide or 150 mg clomipramine. A baselineprincipal component analysis (PCA) was performed to identify atypical symptoms on the com-bined depression scales (Hamilton Depression Scale (HAM-D17) and the Quantitative Scale forAtypical Depression (QSAD)). The primary outcome scale was the subscale HAM-D6 whichcontains the pure items of depression.Results: PCA identified two items with loadings opposite to the other depression items withinHAM-D17 and QSAD, namely increased duration of sleep and increased appetite (atypical neuro-vegetative symptoms). Patients with a positive score at baseline on these items were classified ashaving atypical depression. In total 13 patients were classified as having atypical depression.Within this group of patients 8 received clomipramine and 5 patients received moclobemide. Atendpoint the moclobemide treated patients had a significantly better response than the clomip-ramine treated (P=0.036), effect size 1.42, when using HAM-D6 as outcome. However, in the104 patients classified as having typical depression clomipramine was superior to moclobemide(P=0.034), effect size 0.47.Limitations: The number of patients with atypical neurovegetative symptoms was very small andno placebo arm was included.Conclusions: It is very important to screen for atypical depression (increased duration of sleep/increased appetite) in the acute therapy of patients with major depression. Our results add tothe body of evidence that monoamine oxidase inhibitors are superior to tricyclic antidepressantsin this sub-group of patients.

© 2012 Elsevier B.V. All rights reserved.

Keywords:Atypical depressionMoclobemideClomipramineHamilton Depression Scale

t, Mental Health Centre North Zealand, Dyrehavevej 48, DK-3400 Hillerød, Denmark. Tel.: +45 48 29 32 53;

artment of Psychiatry, Odense University Hospital); Bech P (Department of Psychiatry, Frederiksborg Generalospitalet, University Hospital of Copenhagen); Vestergaard P (Department of Psychiatry, Aarhus Universityksberg Hospital).

ll rights reserved.

http://dx.doi.org/10.1016/j.jad.2012.02.011

mailto:[email protected]

http://dx.doi.org/10.1016/j.jad.2012.02.011

http://www.sciencedirect.com/science/journal/01650327

254 P. Bech et al. / Journal of Affective Disorders 140 (2012) 253–259

1. Introduction

In our struggle to subtype patients with primary depres-sion we focussed on the DSM-IV diagnosis of atypicaldepression, which is considered to be a non-melancholicor non-endogenous subtype for which monoamine oxidaseinhibitors (MAO-Is) are preferred (Bech, 2010b). In thiscontext we referred to the meta-analysis performed byHenkel et al. (2006) which demonstrated that phenelzineobtained an effect size of 0.45 in placebo-controlledtrials on atypical depression. Phenelzine, when comparedto imipramine, still obtained an effect size of 0.27. Inplacebo-controlled trials of antidepressants an effect sizeof 0.40 or higher is considered to be of clinical significance(Bech, 2010a), and in trials where a new antidepressant iscompared to an existing antidepressant, an effect size of0.20 or higher is considered to be of clinical significance(Anderson, 1998).

The symptom profile of patients with atypical depressionwhen compared to patients with typical depression is areversal of the neurovegetative symptoms, i.e. increasedsleep and increased appetite (Henkel et al., 2006). Accordingto DSM-IV, major depression requires at least five of the ninesymptoms of depression where the psychic symptoms ofeither depressed mood or lack of interests are present. Theneurovegetative symptoms of appetite and sleep are consid-ered valid somatic depression items either when decreasedor increased (American Psychiatric Association, 1994). Mosttrials of antidepressants rarely provide information aboutthe score patterns of these neurovegetative symptoms. It isoften shown that up to 30% of unipolar patients with a cur-rent episode of major depression do meet the DSM-IV criteriafor atypical depression (Asnis et al., 1995). In the STAR*Dstudy (Stewart et al., 2010) approximately 20% of the includ-ed patients with DSM-IV major depression were identified assuffering from atypical depression.

In our principal component analysis of the STAR*D base-line dataset (Bech et al., 2011a), we showed that in the firstprincipal component (which was a general factor of depres-sion disability) the item of hypersomnia had a negative factorloading in contrast to all the other items of the Inventory ofDepressive Symptomatology (IDS) (Rush et al., 1986) whichcorresponded to the Hamilton Depression Scale (HAM-D).This finding gave the idea to re-analyse our previous studycomparing the MAO-I moclobemide with clomipramine(DUAG, 1993). In this study we used both the HAM-D/MES(Melancholia Scale) (Bech and Rafaelsen, 1986; Bech et al.,1986) and the Quantitative Scale for Atypical Depression(QSAD), constructed by Larsen et al. (1991), as relevant out-come scales in trials with MAO-Is.

The Henkel et al. (2006) meta-analysis on atypicaldepression concluded that we still have far too few rando-mised controlled trials with moclobemide in this subgroupof primary depression.

The objective of this report, was, therefore, to test by are-analysis of our previous DUAG (1993) study the hypothe-sis in the Henkel et al. analysis that monoamine oxidase in-hibitors (MAO-Is) are superior to tricyclics in the treatmentof patients with atypical depression when comparing moclo-bemide with clomipramine (Bech and Rafaelsen, 1986; Bechet al., 1986).

2. Methods

2.1. Study design

The study was carried out from November 1988 to October1990 in six Danish clinical centres that had also participated inour previous DUAG studies. The patients fulfilled the ICD-8(World Health Organization, 1968) diagnosis of primarydepression. Their score on the HAM-D17 had to be 18 or moreand/or a score of 9 or more on the HAM-D6 (Bech et al.,1986). The patients were then started on a 7-day, single-blindplacebo treatment. A the end of this placeboweek, the patientswere randomly allocated to treatment with a fixed dose ofeither moclobemide 400 mg daily or clomipramine 150 mgdaily, both given in two equal doses per day for 6 weeks(DUAG, 1993).

2.2. Rating scales

The Hamilton Depression Scale (HAM-D17) was usedtogether with the Quantitative Scale for Atypical Depression(QSAD). TheHAM-D17 version used (Bech et al., 1986) containsthe 17 items accepted by Max Hamilton to be the updatedversion of his own scale (Hamilton, 1967). From this scale weemployed the HAM-D6 (Bech et al., 1975) to measure thecore items of depression to be used when assessing the pureantidepressive effect of a drug. These six items are depressedmood, guilt, work and interests, psychomotor retardation,psychic anxiety, and general somatic (pains and tiredness)(HAM-D6).

Because the HAM-D17 is a multidimensional scale, wehave to consider a profile score rather than one single totalor “global” score (Bech, 2011). The specific depression scoreis thus the HAM-D6. Among the remaining items, nine coverthe neurovegetative arousal state (HAM-D9), namely thethree insomnia items (initial, middle, late), agitation, somaticanxiety, decreased appetite, weight loss, decreased sexualinterest, and hypochondriasis (Bech, 2011). The remainingtwo items (HAM-D2) cover lack of insight and suicidalthoughts (Bech, 2011).

In our DUAG study we also included the Melancholia Scale(MES) (Bech and Rafaelsen, 1986) which is based on theHAM-D6 with the addition of such items as decreased sleepin general, motor retardation verbal retardation, concentra-tion difficulties, emotional retardation, and tiredness.

The QSAD items, which were all scored on a Likert scalefrom 0 (not present) to 4 (severely to extremely present)included tension, emotional dependency, increased durationof sleep, and increased appetite (Larsen et al., 1991).

The Newcastle Depression Scale (Bech, 1993; Carney et al.,1965) was used for the diagnosis of endogenous versus non-endogenous depression. The scale contains 10 items, of which‘quality of depression’, ‘weight loss’, ‘previous depressiveepisodes’ and ‘guilt feelings’ are regarded as being in favour ofendogenous depression, whereas ‘personality deviations’,‘psychosocial stressors’, ‘somatic anxiety’, and ‘blaming others’are regarded as being in favour of non-endogenous, or reactive,depression. A score of 6 or more is the cut-off score for endog-enous depression.

Finally, an item on the scoring sheet evaluated whether thepatient had a history of previous episodes with manic states.

255P. Bech et al. / Journal of Affective Disorders 140 (2012) 253–259

2.3. Patient selection

The patients all fulfilled major depression criteria. Inclusioncriteria were a score of 18 or more on the HAM-D17 and/or ascore of 9 on the HAM-D6. Exclusion criteria were age lowerthan 18 years or higher than 70 years, current depressiveepisode duration ofmore than 12months, schizophrenia, para-noid psychosis, mental retardation, organic brain syndrome,chronic alcohol abuse, drug abuse, or serious somatic disease.

2.4. Ethics

Patient consent to participate was obtained on the basis ofwritten and verbal information given to the patients and theirfamily members. The protocol was approved by the regionalEthics Committee and the Danish National Health Service.

2.5. Psychometric analysis

Principal component analysis (PCA) was used at the base-line set (Hotelling, 1933, 1942). The PCA is a purely mathemat-ical analysis, based on intercorrelations between items(Namboodiri, 1984). When two items are correlated, thevariance of each can be divided into two parts, one of whichis common to both items, while the other is specific to eachitem and independent, i.e. orthogonal, to the common varianceand to the other specific variance.When a set of items has beenselected so that the common variance reflects the generaldepressive disability (which is the case for the HAM-D17), allintercorrelations are expected to be more or less positive,resulting in that the first principal component is a generalfactor. In our previous report on the STAR*D study (Bech etal., 2011a), we showed that for the HAM-D17, PCA did indeedidentify a general factor. However, when focussing on theInventory of Depressive Symptomatology (IDS-30) (Rushet al., 1986) we showed that the first principal componentwas a general factor, but that the itemof hypersomniawas neg-atively loaded, implying that this item had no variance in com-mon with the others, i.e. an atypical symptom of depression.

On this background, in the present re-analysis we havefocussed on performing a PCA on the universe of items definedby HAM-D17/MES plus the four extra QSAD items.

If PCA identified some of the QSAD items as atypical, itwas our plan to compare the patients with a positive baselinescore on the atypical items with the group of patients whodid not score on these atypical items.

2.6. Statistical analysis

When comparing the outcome of moclobemide versusclomipramine on the atypically depressed patients we select-ed the HAM-D6 as the primary scale because this subset ofHAM-D items does not contain the QSAD related items.

When comparing moclobemide with clomipramine onthe various outcome scales, we used the last observationcarried forward (LOCF). The LOCF model we have used hasbeen described in detail in the STAR*D analysis (Bech et al.,2011b). Thus, we have excluded any patients without post-randomization observations.

In our analyses we have taken the rather low number of ob-servations in each of the various subtypes of depression into

account. We have therefore used effect size calculations as asupplement to the conventional tests of statistical significance(Hedges and Olkin, 1985). Effect size statistics have to be con-sidered as a measure of clinical significance, using the level of0.40 or higher as acceptable (Bech, 2010a). We have used anonparametric test (Mann–Whitney) for the interpretation ofstatistical significance (Siegel, 1956), accepting a level ofPb0.05, two sided. The Mann–Whitney test was performedon change score as regards the various rating scales from base-line in order to increase the correspondencewith the effect sizestatistics. When testing the baseline scores themselves, theMann Whitney was obviously performed on these scores.

For all these statistical analyses we have used the SAS(version 9.00, 2002) programme.

3. Results

Fig. 1 shows the flow diagram. In total 141 patients werescreened and 117 patients were randomised after one-weeksingle-blind placebo therapy. In the DUAG (1993) publicationtwo patients were retrospectively not considered for analysisbecause both had scored just below 18 on the HAM-D17 orjust below 9 on the HAM-D6 at randomisation. Of the 117patients 58 received moclobemide and 59 patients receivedclomipramine (Fig. 1). The reasons for exclusion of the 24patients after the one-week single-blind placebo period arelisted in Fig. 1. The term response covers a 50% or higher reduc-tion on the HAM-D17. During the planned six-week double-blind trials 20 patients dropped out on moclobemide and 12patients on clomipramine. In total 73% completed the trial(66% formoclobemide and 79% for clomipramine). One patienton moclobemide and one patient on clomipramine developedmania. Neither of these patients had atypical depression.

On moclobemide, 11 patients worsened/had insufficienteffect against 4 patients on clomipramine, and on moclobe-mide 4 patients developed severe side effects (one patientdue to skin rash, one patient due to clinical signs of myocardialinfarction which, however, was not confirmed, one patientwithmild hypertension, and one patientwith orthostatic hypo-tension) versus 6 patients on clomipramine (all six patientsdue to orthostatic hypotension and dizziness) (Fig. 1).

Table 1 shows the factor loadings on the first principal com-ponent identified by PCA. The eigenvaluewas 4.17. In total ninefactors with eigenvalues above 1 were identified. As shown inTable 1, the items within the HAM-D17 with MES plus theQSAD items were all positively loaded apart from the items ofincreased duration of sleep and increased appetite, which hadnegative loadings.

Table 2 shows the outcome of the study for all 117 patients(clomipramine=59 and moclobemide=58) on the HAM-Dprofile when taking into account both (A) the specific HAM-D6 scale; (B) the HAM-D9 with the arousal items; (C) theHAM-D2 covering suicidal behaviour, and finally the totalHAM-D17 (A+B+C). Clomipramine was statistically superiorto moclobemide on the total (ABC) HAM-D17 (Pb0.05); how-ever, this superiority was explained by the anti-arousal effect(B, or HAM-D9) as indicated in Table 2. The effect size statisticsconfirmed that a level of clinical significance (effect size 0.40 orhigher) was obtained for the total (ABC) HAM-D17 as for theHAM-D9.


Table 3 shows the 13 patients or 11.1% who had positivescores on the atypical items of increased sleep and/or increasedappetite. In this atypical depression group, 8 patients receivedclomipramine and 5 patients received moclobemide. On theNewcastle 1965 scale approximately 65% in both groups wereclassified as endogenously depressed. The two patients retro-spectively excluded in the DUAG (1993) publication are notto be found among the 13 patients in Table 3.

After 3 weeks of therapy the difference between moclobe-mide and clomipramine was statistically significant (Pb0.027),and this was sustained to week 6 (Pb0.036). As early as after1 week of treatment effect size statistics showed the superiorityof moclobemide over clomipramine, i.e. effect sizes greater

Patient flow

Excluded during the planned trials

enimarpimolCedimebolcoM

Worsening/unsatisfactory effect

N = 11 N = 4

1=N1=N

Severe side effects N = 4 N = 6

Non-compliance N = 4 N = 1

21=N02=NlatoT

Screened N = 141

Excluded in the placebo wash-out week N = 24

Responded N = 6

Worsening/suicidal N = 4

Wanted to stop N = 4

Blood analysis problems N = 9

Changed diagnosis N = 1

Moclobemide: 58

Clomipramine: 58

Randomised N = 117

)%66(83=N:edimebolcoM

Completers

)%97(74=N:enimarpimolC

Mania

Fig. 1. Patient flow.

Table 1Principal component analysis (PCA) on the baseline scores for the 117 patientsat the time of randomization. Factor loadings for the first principal component.Baseline DUAG-3 N=117.

Item Factor loadings

1 Depressed mood 0.652 Guilt 0.383 Suicide 0.504 Insomnia early 0.435 Insomnia middle 0.176 Insomnia late 0.267 Work and interests 0.558 Motor retardation 0.609 Agitation 0.1210 Psychic anxiety 0.3711 Somatic anxiety 0.1512 Gastroinst. 0.4313 Somatic, general 0.0114 Sexual interest 0.0515 Hypochondriasis 0.1716 Insight 0.2717 Weight loss 0.2418 Sleep, general 0.4119 Motor retardation 0.6520 Verbal retardation 0.6621 Concentration 0.5122 Emotional retardation 0.4423 Tiredness 0.2124 Tension 0.3725 Emotional 0.2426 Increased duration of sleep −0.1227 Increased appetite −0.07

than 0.40. Among the 13 patients with atypical neurovegetativesymptoms, one patient droppedout due to side effects onmoclo-bemide and two patients on clomipramine.

Table 4 shows the outcome on HAM-D6 in the typicallydepressed patients (i.e. without hypersomnia and/or increasedappetite). After 6 weeks of therapy clomipramine was statisti-cally superior tomoclobemide (Pb0.034). The effect size statis-tics confirmed this with a level greater than 0.40. In both thetypical depression group and in the atypical depressiongroup, approximately 70% scored endogenous on the Newcas-tle (1965) Scale.

Among the 58 patients receiving clomipramine, seven pa-tients had previously developed manic states (12.1%); andamong the 56 patients receiving moclobemide, five had previ-ously developedmanic states (8.9%). Among these 12 bipolar de-pressed patients none had atypical neurovegetative symptoms.

Table 5 shows the outcome on the HAM-D6 for thesebipolar I depressed patients who all scored endogenouslydepressed on the Newcastle (1965) scale.

The results in Table 5 showed that the superiority of clomip-ramine over moclobemide did not reach the level of statisticalsignificance (P=0.059), but after 1 week of therapy the effectsizes were already above the level of clinical significance(0.40). One patient in the moclobemide group and one patientin the clomipramine group developed mania, this was consid-ered to be a side effect.

4. Discussion

This re-analysis of the DUAG (1993) study comparing theeffect of the tricyclic antidepressant clomipramine with thesecond generation MAO-I moclobemide has shown the

Table 4Patients without hypersomnia and/or increased appetite (typical depression).(N=104). Outcome measure: HAM-D6, LOCF.

Clomipramine(N=51)

Moclobemide(N=53)

Effectsize

MannWhitney

Baseline 12.6 (2.3) 12.3 (2.5) 0.353Week 1 11.3 (2.3) 11.0 (3.2) 0.11 0.455Difference 1.3 (2.1) 1.3 (2.0) 0.00 0.795Week 2 10.2 (2.5) 10.4 (3.6) 0.06 0.621Difference 2.4 (2.8) 1.9 (2.8) 0.18 0.564Week 3 8.8 (2.7) 9.1 (4.1) 0.09 0.600Difference 3.8 (3.3) 3.3 (4.0) 0.14 0.524Week 4 7.9 (3.2) 8.9 (4.5) 0.25 0.145Difference 4.7 (3.9) 3.4 (4.4) 0.31 0.119Week 5 7.5 (3.4) 8.6 (4.6) 0.27 0.207Difference 5.1 (4.3) 3.7 (4.6) 0.31 0.213Week 6 6.5 (4.1) 8.6 (4.9) 0.46 0.027Difference 6.1 (5.0) 3.8 (4.7) 0.47 0.034

Table 2The ABC profile of the HAM-D17, using the LOCF approach.

Clomipramine(N=59)

Moclobemide(N=58)

Effectsize

MannWhitney

A+B+C HAM-D17

Baseline 23.3 (4.5) 23.2 (5.4) 0.840Endpoint 11.9 (7.2) 15.7 (9.1) 0.47 0.018Difference 11.4 (8.8) 7.5 (8.3) 0.45 0.045

A HAM-D6 (depression)Baseline 12.5 (2.2) 12.5 (2.5) 0.798Endpoint 6.9 (4.0) 8.5 (4.7) 0.38 0.046Difference 5.6 (4.9) 3.9 (4.6) 0.35 0.124

B HAM-D9 (arousal)Baseline 9.2 (2.9) 9.1 (3.4) 0.677Endpoint 4.5 (3.6) 6.1 (4.3) 0.40 0.024Difference 4.7 (4.3) 3.0 (4.1) 0.41 0.048

C HAM-D2 (suicide)Baseline 1.6 (1.0) 1.6 (1.2) 0.948Endpoint 0.6 (0.9) 1.1 (1.4) 0.48 0.025Difference 1.0 (1.1) 0.5 (1.5) 0.39 0.084


importance of subtyping depressed patients into typical ver-sus atypical depression on the basis of absence versus pres-ence of the neurovegetative symptoms of increased sleepand increased appetite.

The supplemental use of effect size statistics to indicateclinical significance when comparing clomipramine withmoclobemide confirmed the test for statistical significancewhen the number of observations was acceptable. The smallnumber of observationswhen comparing subgroups illustratedthe importance of using effect size statistics. In this connectionwe should emphasize that we have used the unbiased versionin the calculation of effect sizewhich controls for small numberof observations (Bech, 2007; Hedges and Olkin, 1985).

In total 11.1% of the included patients fulfilled our criteriafor atypical depression, i.e. patients with atypical neurovegeta-tive symptoms. This is onlymarginally higher than the percent-age found in the GENDEP (Genome-based Therapeutic Drugsfor Depression) by Uher et al. (2011). However, in this studyno statistically significant difference in outcome on the HAM-D17 was obtained between typical and atypical patients whennortriptyline was compared to escitalopram.

Table 3Patients with hypersomnia and/or increased appetite (atypical depression).(N=13). Outcome measure: HAM-D6, LOCF.

Clomipramine(N=8)

Moclobemide(N=5)

Effectsize

MannWhitney


The ABC profile of the HAM-D (Table 2), in which theneurovegetative symptoms of decreased sleep and decreasedappetite/weight loss cover the unspecific arousal syndrome(Bech, 2011) demonstrated the outcome also released byDUAG (1993), namely that clomipramine is superior to moclo-bemide. However,when focussing on the patientswith atypicaldepression by the criteria obtained with principal componentanalysis (increased sleep and increased appetite), we wereable to demonstrate (Table 3) that moclobemide was superiorto clomipramine. In the group suffering from typical depression(Table 4) the superiority of clomipramine over moclobemidewas sustained. This was also the case for the subgroup of bipo-lar depressed patients (Table 5).

Throughout this re-analysis of the DUAG (1993) results, wehave used the HAM-D6 subscale as outcome measure whencomparing typical versus atypical depression, as this subscaleincludes specific depression symptoms outside the range ofthe arousal items, which should be considered as the unspecificstress reaction (Bech, 2011).

When comparing typical versus atypical depression weneed to demonstrate that even the atypical patients are indeeddepressed. Therefore the use of the specific depression scale(HAM-D6) is also important at baseline assessment. Table 3shows that both the typically and the atypically depressed pa-tients scored approximately 12.5 on the HAM-D6 at baseline,

Table 5Patients with bipolar 1 depression. (N=12). Outcome measure HAM-D6,LOCF.

Clomipramine(N=7)

Moclobemide(N=5)

Effectsize

MannWhitney



this indicates both a rather marked state of depression and ex-plains approximately 54% of the variance with reference to thebaseline score on HAM-D17.

Another way to confirm that patients with atypical depres-sion belong within the category of primary depression (Bech,2010b) is the identification of an endogenous symptompatternon the Newcastle (1965) depression scale and the identifica-tion of bipolar depression. In the atypical depression groupdiagnosed in this report, the percentage of endogenous depres-sion as well as of bipolar depression was similar in atypicalversus typical depression.

Our results seem to support the findings in themeta-analysisby Henkel et al. (2006), that the first generation MAO-I phenel-zine is superior to non-MAO-I antidepressants in the treatmentof atypical depression. However, we still need more trials onthe second generation MAO-Is to accept the label of atypicaldepression for moclobemide. We also need to investigate therecommended dose range of moclobemide from 300 to 600 mgdaily (Bonnet, 2003).

As discussed elsewhere, atypical depression is within DSM-IV considered as a non-melancholic subtype (Bech, 2010b).However, when restricting the term atypical depression to in-verse neurovegetative symptoms we are operating within thecategory of melancholic or endogenous depression (Bech,2010b; Vanggaard, 1976). Patients with bipolar depressionare to be placed within the category of melancholia and theyoften demonstrate atypical symptoms as shown by Mitchellet al. (2001). On this background a proposed definition of bipo-lar spectrum disorder includes the items of atypical depressionsymptoms (Ghaemi et al., 2002). In our study the patients withatypical neurovegetative symptoms were predominantly en-dogenously depressed as measured by the Newcastle Diagnos-tic Depression Scale, but had no previous episodes of mania.

As discussed by Goodwin (2012), we still have very limiteddata from randomised clinical trials to evaluate the efficacy ofantidepressants in bipolar depression. When referring toantidepressants in this respect, it is the group of reuptake-inhibitors and not monoamine oxidase inhibitors we are refer-ring to. Our results confirm the results of the Henkel et al.(2006) meta-analysis: that monoamine oxidase inhibitors aresuperior to reuptake inhibitors such as clomipramine inpatients with atypical neurovegetative symptoms. We needto investigate the role of monoamine oxidase inhibitors suchas moclobemide in bipolar depression with atypical neurove-getative symptoms. Although 400 mg moclobemide, as usedin this study, might be found a low dose, we feel that in bipolardisorder dose–response trials should be considered, within therecommended dose range of 300 to 600 mg daily.

Role of funding sourceNo funding source involved.

Conflict of interestPer Bech: none to declare.Kurt B. Stage: none to declare.Jens Knud Larsen: none to declare.Per Vestergaard: none to declare.Lars F. Gram: none to declare.

AcknowledgementsNone.

References

American Psychiatric Association, 1994. The Diagnostic and Statistical Manualof Mental DisordersFourth Edition. American Psychiatric Association,Washington D.C.. (DSM-IV).

Anderson, I.M., 1998. SSRI's versus tricyclic antidepressants in depressed inpa-tients: a meta-analysis of efficacy and tolerability. Depression and Anxiety7 (Suppl 1), 11–17.

Asnis, G.M., McGinn, L.K., Sanderson, W.C., 1995. Atypical depression: clinicalaspects and noradrenergic function. The American Journal of Psychiatry152, 31–36.

Bech, P., 1993. Rating scales for psychopathology, health status and quality oflife. A Compendium on Documentation in Accordance with the DSM-III-R and WHO Systems. Springer, Berlin.

Bech, P., 2007. Dose–response relationship of pregabalin in patients withgeneralized anxiety disorder. A pooled analysis of four placebo-controlledtrials. Pharmacopsychiatry 40, 163–168.

Bech, P., 2010a. Is the antidepressive effect of second-generation antidepressantsa myth? Psychological Medicine 40, 181–186.

Bech, P., 2010b. Struggle for subtypes in primary and secondary depression andtheirmode-specific treatment or healing. Psychotherapy and Psychosomatics79, 331–338.

Bech, P., 2011. The ABC profile of the HAM-D17. Revista Brasileira de Psiquiatria33, 109–110.

Bech, P., Rafaelsen, O.J., 1986. The Melancholia Scale: development, consistency,validity and utility. In: Ban, T., Sartorius, N. (Eds.), Assessment of Depression.Springer, Berlin, pp. 259–269.

Bech, P., Gram, L.F., Dein, E., Jacobsen, O., Vitger, J., Bolwig, T.G., 1975. Quantitativerating of depressive states. Acta Psychiatrica Scandinavica 51, 161–170.

Bech, P., Kastrup, M., Rafaelsen, O.J., 1986. Mini-compendium of rating scales forstates of anxiety, depression, mania, and schizophrenia with correspondingDSM-III syndromes. Acta Psychiatrica Scandinavica. Supplementum 326,1–37.

Bech, P., Fava, M., Trivedi, M.H., Wisniewski, S.R., Rush, A.J., 2011a. Factorstructure and dimensionality of the two depression scales in STAR*Dusing level 1 datasets. Journal of Affective Disorders 132, 396–400.

Bech, P., Fava, M., Trivedi, M.H., Wisniewski, S.R., Rush, A.J., 2011b. Outcomes onthe pharmacopsychometric triangle in bupropion-SR vs. buspirone augmen-tation of citalopram in the STAR*D trial. Acta Psychiatrica Scandinavica.doi:10.1111/j.1600-0447.2011.01791.

Bonnet, U., 2003. Moclobemide: therapeutic use and clinical studies. CNSDrug Reviews 9, 97–140.

Carney, M.W., Roth, M., Garside, R.F., 1965. The diagnosis of depressivesyndromes and the prediction of E.C.T. response. The British Journal ofPsychiatry 111, 659–674.

DUAG, 1993. Moclobemide: a reversible MAO-A-inhibitor showing weakerantidepressant effect than clomipramine in a controlled multicenterstudy. Danish University Antidepressant Group. Journal of AffectiveDisorders 28, 105–116.

Ghaemi, S.N., Ko, J.Y., Goodwin, F.K., 2002. “Cade's disease” and beyond: misdiag-nosis, antidepressant use, and a proposed definition for bipolar spectrumdisorder. Canadian Journal of Psychiatry 47, 125–134.

Goodwin, G.M., 2012. Bipolar depression and treatment with antidepressants.The British Journal of Psychiatry 200, 5–6.

Hamilton, M., 1967. Development of a rating scale for primary depressiveillness. The British Journal of Social and Clinical Psychology 6, 278–296.

Hedges, L.V., Olkin, I., 1985. Statistical Methods for Meta-Analysis. AcademicPress, Orlando, FL.

Henkel, V., Mergl, R., Allgaier, A.K., Kohnen, R., Möller, H.J., Hegerl, U., 2006.Treatment of depression with atypical features: a meta-analytic approach.Psychiatry Research 141, 89–101.

Hotelling, H., 1933. Analysis of a complex of statistical variables with principalcomponents. Journal of Educational Psychology 24, 417–441.

Hotelling, H., 1942. Rotations in psychology and the statistical revolution.Science 95, 504–507.

Larsen, J.K., Gjerris, A., Holm, P., Anderson, J., Bille, A., Christensen, E.M.,Hoyer, E., Jensen, H., Mejlhede, A., Langagergaard, A., 1991. Moclobemidein depression: a randomized, multicentre trial against isocarboxazideand clomipramine emphasizing atypical depression. Acta PsychiatricaScandinavica 84, 564–570.

Mitchell, P.B., Wilhelm, K., Parker, G., Austin, M.P., Rutgers, P., Malhi, G.S., 2001.The clinical features of bipolar depression: a comparison with matchedmajor depressive disorder patients. The Journal of Clinical Psychiatry 62,212–216.

Namboodiri, K., 1984. Matrix Algebra. An Introduction. SAGE Publications,London.

Rush, A.J., Giles, D.E., Schlesser, M.A., Fulton, C.L., Weissenburger, J., Burns, C.,1986. The Inventory for Depressive Symptomatology (IDS): preliminaryfindings. Psychiatry Research 18, 65–87.

http://dx.doi.org/10.1111/j.1600-0447.2011.01791


Siegel, S., 1956. Nonparametric Statistics for the Behavioural Sciences.McGraw Hill, New York.

Stewart, J.W., McGrath, P.J., Fava, M., Wisniewski, S.R., Zisook, S., Cook, I.,Nierenberg, A.A., Trivedi, M.H., Balasubramani, G.K., Warden, D., Lesser, I.,John Rush, A., 2010. Do atypical features affect outcome in depressed outpa-tients treated with citalopram? The International Journal of Neuropsycho-pharmacology 13, 15–30.

Uher, R., Dernovsek, M.Z., Mors, O.E.A., 2011. Melancholic, atypical and anxiousdepression subtypes and outcome of treatment with escitalopram andnortriptyline. Journal of Affective Disorders 132, 112–120.

Vanggaard, T., 1976. Atypical endogenous depression. Acta PsychiatricaScandinavica. Supplementum 7–56.

World Health Organization, 1968. International Classification of Diseases(ICD-8). World Health Organization, Geneva.

Documents

The predictive validity of atypical neurovegetative depressive symptoms identified by the first principal component in the DUAG trial of moclobemide versus clomipramine