Original Author: Paul Perry, Ph.D, BCPP
Latest Reviser: Paul Perry, Ph.D, BCPP, Brian C. Lund, Pharm.D,
BCPP
Creation Date: 1996
Last Revision Date: December 2000
Peer Review Status: Internally Peer Reviewed
Antidepressants are used for the treatment of affective or mood disorders which have as one of their components a depressive syndrome. Depressive syndrome variants include organic induced depressions, bipolar depression, schizoaffective depressions, and unipolar depressions (Winokur, 1991). With such divergent presentations of major depression, it may be overly simplistic to expect all antidepressant drug classes to be equally effective in their treatment.
In terms of DSM criteria, the diagnosis of endogenous depression is comparable to major depressive disorder with melancholic features. As an example of this diagnostic equality, Zimmerman et la (1984) reviewed the charts of 271 depressed inpatients who met DSM-III criteria for major depressive disorder, 110 (41%) were melancholic. All 110 melancholics met the Research Diagnostic Criteria for endogenous depression (Spitzer et al., 1978). Stage et al (1998) has demonstrated that the symptom presentation of depressed inpatients differs remarkably from depressed outpatients. Three inpatient studies of depression treatment efficacy were contrasted to three outpatient studies. The inpatient (n=352 studies concluded that the tricyclic antidepressants, clomipramine, was more effective than the SSRIs, citalopram and paroxetine, and the MAOI, moclobemide. The outpatient (n=581 studies concluded that there was no difference in effectiveness between the tricyclic antidepressant, imipramine and clomipramine, and the SSRIs, citalopram and paroxetine, and the MAOI, moclobemide. These studies utilized the diagnoses of endogenous and nonendogenous depression. According to DSM-IV criteria, these diagnoses correspond to major depression with melancholic features and major depression without melancholic features. In total, 76% of the inpatients and 40% of the outpatients had melancholic/endogenous depression. The inpatients were older, had higher scores on the Hamilton and Newcastle depression scales and had more symptoms of guilt, nihilistic delusions, retardation, and weight loss. The outpatients reported more psychological stressors and anxiety. Thus the important point to glean from these two studies is that a significant percentage of depressed inpatients meet the criteria for major depression with melancholic features regardless of whether they are in the US or Europe (Denmark).
The obvious question to consider is whether there is any difference in effectiveness between the tricyclic antidepressants and the SSRI antidepressants in the treatment of MDD with melancholic features. In 1986, a controversial difference between the effectiveness of TCAs and SSRIs was noted in a study conducted by the Danish University Antidepressant Group (DUAG) (1986). Endogenous and nonendogenous depression (Newcastle II Scale) was diagnosed in 114 inpatients with Hamilton Depression Scale (HAMD17) scores of >18. As previously mentioned, endogenous depression is comparable to major depressive disorder with melancholic features (Spitzer et al., 1978). The essential feature of this specifier is a loss of interest in most activities, or lack of reactivity to usually pleasurable stimuli, i.e., anhedonia (DSM-IV 1994). Following a one-week placebo lead-in patients were randomized to either a fixed dose of the TCA, clomipramine 150 mg/d or the SSRI, citalopram 40 mg/d for 5 weeks. The drug response for endogenous depression differed in that a remission (HAMD17 < 7) was more likely in the clomipramine treated patients (62%) than citalopram patients (34%). This response pattern favoring the TCAs over the SSRIs was also apparent for the non-endogenous diagnosed patients. Importantly, there were no response differences between the drugs in the distribution of partial responders (HAMD17 = 8-15) for either the endogenous depression or the non-endogenous depression groups. Additionally, there was no difference in response rates if the partial and full responses are totaled. The authors concluded that the SSRIs were less effective than the TCAs in the treatment of endogenous depression. If the TCAs are indeed more effective in the treatment of major depression with melancholic features this would constitute significant savings for the health care systems’ drug budget.
The data summary which follows summarizes the concordant replication studies of the DUAG findings. Additional studies that have attempted to answer this question are discussed. However, they provided insufficient data in their results from which any positive or negative conclusion can be formed.
Concordant Replication Studies
Following a 7-day placebo lead-in period, Laakmann et al (1988) randomized 105 patients with endogenous depression (ICD-9: 296.1/296.3) and a HAMD > 17 to either the TCA, amitriptyline 50-150 mg/d or the SSRI, fluoxetine 20-60 mg/d for 5 weeks. Overall, no statistically significant difference in effectiveness was noted between amitriptyline and fluoxetine. However, the patients' depressions were further subtyped as being either "retarded" or "vitally disturbed" or "anxious-agitated". The distribution of responders and nonresponders was not presented. No difference was observed in the retarded depression group HAMD scores, although the trend favored amitriptyline. There were only 14 fluoxetine and 11 amitriptyline patients included in this retarded depression sample. A power analysis (beta = 0.8, alpha = 0.05) predicted a sample size of 31 patients per group would have demonstrated a significant difference. Despite this small sample size, there was a significant difference in the retarded depression group's Self Rating Depression Scale scores that favored amitriptyline over fluoxetine. Patients were also subdivided in accordance with the severity of their illness on the Clinical Global Impression Scale as being either mildly ill, moderately ill, or severely-extremely ill. Importantly, severity of illness did not segregate the treatment response as did the endogenous-nonendogenous dichotomy. It is often suggested that endogenous or melancholic depressions are more severe forms of depression. These data do not support this contention.
The DUAG (1990) repeated their benchmark study except they substituted the SSRI, paroxetine for citalopram. Using the identical protocol as the initial study, 120 inpatients were randomized to either a fixed dose of clomipramine 150 mg/d or paroxetine 30 mg/d for 6 weeks. For endogenous depression, a complete remission (HAMD17 < 7) was more likely in the clomipramine treated patients (56%) than the paroxetine patients (25%). The partial response rates were similar at 30% and 25% for clomipramine and paroxetine, respectively. However, the 50% non-response rate for paroxetine was greater than the 14% rate for clomipramine. As in the first study, the response pattern favoring clomipramine was also consistent for the non-endogenous diagnosed patients. The complete, partial, and no response rates for clomipramine were 60%, 30% and 10% respectively while the respective rates for paroxetine were quite discouraging at 12%, 44$, and 44%, respectively. Both DUAG studies used a fixed dose of drug. Because of the greater risk of adverse effects with TCAs, this is an important design factor since there is a real chance in flexible dose studies that patients treated with the TCAs could complete the studies on subtherapeutic doses of the drug resulting from attempts to avoid adverse effects.
Tignol et al (1992) performed a meta-analysis of the manufacturer’s database involving paroxetine treatment of melancholic depressions (DSM-III). The analysis included 178 patients treated with paroxetine and 66 patients treated with placebo. The paroxetine response data were remarkably consistent with the two DUAG studies. Paroxetine produced a greater response than placebo on two HAMD parameters, i.e., HAMD <10 and > 50% decrease. More importantly and similar to the previous studies, only 31% of the paroxetine patients had a HAMD <10 while 15% of the placebo treated patients had a HAMD <10. Once again study data demonstrate a less than robust response for an SSRI in the treatment of melancholic depression.
Roose et al (1994) retrospectively contrasted the treatment responses of the TCA, nortriptyline and fluoxetine in a group of 45 depressed melancholic (DSM-III or DSM-III-R) geriatric inpatients with severe cardiovascular disease, i.e., an ejection fraction of <50%, >10 premature ventricular contraction per hour, and/or a QRS interval of >0.10 sec. Following a two-week placebo lead-in patients were treated by non-randomized assignment for seven weeks with either nortriptyline 1 mg/kg/d with the blood level adjusted to 50-150 ng/ml or fluoxetine 20 to 60 mg/d. As in the previous studies, the findings favored the TCA over the SSRI. Including all the dropouts, 63% (20/32) of the nortriptyline group and only 8% (1/13) (of the fluoxetine group had a final HAMD21 score of <8. Despite the retrospective nature of these data, the clinically relevant findings are remarkably similar to the previous four studies.
Mulsant et al (1999) conducted a controlled trial contrasting the effectiveness of a 6-week trial of paroxetine 20-30 mg/d to nortriptyline 50-150 ng/ml in 80 depressed in- and out-patient (DSM-IV) elderly (mean=75.0 years). Rates of response (HAMD <10) were compared in the melancholic patients. Predictably the percentage of patient responding to nortriptyline (13/21, 62%) exceeded paroxetine (10/23, 43%) even though the difference was not significant (p = 0.25).
Heiligenstein et al (1994) compared the efficacy of fluoxetine (20 mg/d) to placebo in the treatment of 52 outpatients with major depression (DSM-III-R) with melancholia in an 8-week trial. Fluoxetine was more effective than placebo within the melancholic subtype. Based on 48 patients who completed at least three weeks of treatment, remissions (HAMD17 < 6) occurred in 48% (10/21) of the fluoxetine patients and only 11% (3/27) of the placebo-treated patients. This is the most robust treatment response for a SSRI in a melancholic population. It was also an outpatient population that only required a 15 on the HAMD17 for inclusion and whose mean HAMD17 at baseline was approximately 22. These melancholic patients were possibly less severely ill than the patients in the other studies.
Mendels et al (1999) contrasted the effectiveness of a 4-week trial of citalopram 20-80 mg/d to placebo in the treatment of major depression in an outpatient population. Most of the 180 patients had a diagnosis of melancholic depression (n=165). The rate for a full remission cannot be ascertained from the HAMD data analysis. However, overall 81% of the citalopram patients and 47% of the placebo group exhibited a partial response according to the CGI rating (much or very much improved). The data are less than overwhelming since this was an outpatient population with a remarkably high placebo response rate.
Other Studies
One argument that attempts to discount the above findings involves the hypothesis that severe depressions are synonymous with endogenous depression while mild-moderate depression are more likely to be non-endogenous. However, this hypothesis is rejected by the TCA versus SSRI antidepressant treatment literature. As suggested by Laakman (1988), as well as a number of other studies, there is no difference between the efficacy of TCAs and SSRIs in severe depressions. However, there are problems with these studies that preclude their use in the literature analysis of the melancholic depression treatment question. Beasley et al (1993) evaluated the treatment response to imipramine and fluoxetine in 118 severely depressed inpatients with major depression (DSM-III). HAMD baseline scores were 27 for imipramine and 28 for fluoxetine. There was no difference in the remission rates (HAMD21 < 7) produced by fluoxetine (21%) and imipramine (34%) following a six-week trial. Unfortunately the melancholic patients were not separated out in the analysis. Bowden et al (1993) found no response difference between the TCA, desipramine 100-300 mg/d (68%), and fluoxetine 20-60 mg/d (64%), in the treatment of 58 inpatients and outpatients with severe major depression (DSM-III-R). Mean baseline HAMD scores were 26 for desipramine and 25 for fluoxetine. Response was defined as a > 50% decrease in the HAMD. Again, the melancholic patients were not analyzed separately. Stuppaeck et al (1994) in a 6-week multi-center trial found no difference in the effectiveness of paroxetine 30-50 mg/d and amitriptyline 150-250 mg/d in the treatment of 134 depressed melancholic patients (DSM-III). However, the definition of response on the HAMD21 data was inconsistent with previous studies. Response was defined as either a 50% decrease in the baseline HAMD21 or as a score < 14. The percentages of responders for both criteria were nearly identical. The response rates for the HAMD > 50% decrease and HAMD < 14 for amitriptyline were 70% and 64%, respectively and 63% and 66% for fluoxetine, respectively. Thus if these findings are put in the context of the DUAG (1986, 1990) studies, they concluded for melancholic (endogenous) depressed patients there was no difference in the distribution of the partial and complete responders compared to the nonresponders for the treatment of depression with a TCA and an SSRI. Indeed, this finding is in agreement with the first DUAG (1986) in which it there was no difference in the complete and partial response rates between clomipramine (70%) and citalopram (66%). However, the second DUAG (1990) trial did find a difference that favored clomipramine (86%) over paroxetine (50%).
Discussion
Clomipramine is a significantly more potent inhibitor of serotonin reuptake than the other TCAs (Nelson et al., 1989). Coupling its significant norepinephrine agonist activity with the potent serotonergic agonist activity suggests that clomipramine is an atypical TCA. It could be argued of the TCAs, only clomipramine is more effective than SSRIs in the treatment of melancholia. In an open trial, Nelson et al (1991) treated 14 inpatients with major depressive disorder (DSM-III-R) with a combination of desipramine and fluoxetine. The desipramine dose was adjusted into the therapeutic range to take into account the inhibitory effect of fluoxetine on desipramine metabolism. After 4 weeks of treatment, the combination therapy resulted in a 71% response rate in contrast to a 14% response rate in 42 similar patients treated with desipramine alone. Thus the TCA was only effective after it serotonin agonist activity was augmented by fluoxetine. Thus it could be hypothesized of the TCAs only clomipramine has enough serotonin agonist activity to be effective in difficult to treat depressed patients. However, the idea that clomipramine is unique is discounted by the Roose (1994) study that found nortriptyline more effective than fluoxetine in elderly melancholic patients. Thus age as well might confound these data. Tollefson and Holman (1993) contrasted the remission rates of 671 elderly (> 60 years) depressed (DSM-III-R compatible) patients randomized to either fluoxetine or placebo. At four weeks, the remission HAMD <7) rates were a disappointing 21% for fluoxetine and 13% for placebo as were the partial response rates (HAMD 50% decrease) of 36% for fluoxetine and 27% for placebo. Thus these data suggests that SSRIs produce modest responses rates in elderly patient regardless of whether they have a melancholic diagnosis.
Venlafaxine and Nefazodone
Several studies have contrasted the effectiveness of the venlafaxine and nefazodone against fluoxetine, imipramine, and placebo in the treatment of melancholic depression.
In a 6-week double-blind multicenter study, Clerc et al (1994) randomized 67 inpatients diagnosed with major depression with melancholia to either venlafaxine 200 mg/d or fluoxetine 40 mg/d. The HAMD scores at weeks 4 and 6 favored venlafaxine. The final HAMD change scores for venlafaxine (29+5 ->11+10) reflected the drug’s superiority over fluoxetine (30+4->17+12). A response rate defined as a 50% decrease in HAMD, found 73% of the patients responding to venlafaxine versus a 50% response rate for fluoxetine. Despite these suggestive positive findings the investigators did not report the percentage of patients with HAMD scores of < 7. However, in a later article, Thase et al (2001) reported a response rate (HAMD < 7) of 55% for venlafaxine and only 26% for fluoxetine. The intriguing part of these remission rates are that they are similar to those observed in the TCA efficacy studies. The multicenter placebo-controlled double-blind study conducted Guelfi et al (1995) lends credibility to this concern. Ninety-three inpatients diagnosed with major depression with melancholic features were randomized to either placebo or venlafaxine (mean = 350 mg/d) for a 4-week trial. The venlafaxine mean HAMD scores decreased from 29 to 14 whereas the placebo treated patients had their scores decrease only from 29 to 24. The most important finding, the percentage of patients with HAMD scores < 7 found a 26% rate for venlafaxine and 13% rate for placebo. Thus the venlafaxine full response rate was similar to the SSRIs rather than the TCAs. Venlafaxine was compared to imipramine in a 6-week double-blind multicenter study (Benkert 1996). Venlafaxine was rapidly titrated up to 375 mg/day and maintained at this dose from days 5-14, then decreased to 150 mg/day for the remaining 4 weeks. Imipramine was increased over five days to 200 mg/day and maintained for the remainder of the study. A response of >50% reduction in HAMD scores at 6 weeks was observed in 52% of the venlafaxine treated patients, and 59% in patients receiving imipramine. The endpoint HAMD scores were not given and there was no indicator of remission rates. Tzanakaki (2000) conducted a muticenter randomized controlled trial in which venlafaxine 225 mg/d was contrasted to fluoxetine 60 mg/d to treat 93 inpatients with MDD with melancholic features. There were 12 dropouts in each treatment group. The response rate was similar according to the HAMD criteria of (<7) for venlafaxine (41%, 22/55) and fluoxetine (36%, 19/54) Additionally, a statistically similar number of patients responded to venlafaxine (51%, 28/55) according to the CGI criterian of very much improved in contrast to fluoxetine (32%, 17/54). For practical purposes however, these results do not deviate from the consistently observed superiority of the tricyclics over the other antidepressants in the treatment of the melancholic depressions.
Marcus et al (1996) provided an indication as to the effectiveness of nefazodone in the treatment of melancholic depression. The authors analyzed data on 664 patients with either severe, melancholic, or recurrent depression from four randomized placebo-controlled double-blind studies that were 6 to 8 weeks in duration. Fifty-five percent (364) patients were diagnosed as melancholic. The mean dose of nefazodone was 379 mg/d versus 166 mg/d for the comparator drug imipramine. For both nefazodone and imipramine the mean HAMD score decreased from 26 to 13 while for placebo the score dropped from 26 to 17. According to the clinical global impression improvement score a rating of very much or much improved was observed in 60% of the imipramine patients, 58% of the nefazodone patients, and 41% of the placebo patients. The two obvious concerns here are the high placebo response rate and not knowing what percentage of patients achieved the "gold standard" HAMD score of < 7.
Electroconvulsive Therapy
This review would not be complete unless a reference to the effectiveness of ECT in the treatment of melancholia was included. Janakiramaiah et al (1998) treated 80 melancholic patients (DSM-III-R) with one or three times weekly bilateral electroconvulsive treatments. A response and termination of the treatments was defined a priori as a HAMD score of <7. At the end of 4 weeks85% (17/20) of the three times weekly group had remitted versus 55% (11/20) of the once weekly groups. As one might have guessed, despite the effectiveness of the TCAs in the inpatient treatment of major depression with melancholic features, ECT effectiveness still probably is greater.
Summary
Comparative HAMD data from controlled trials lead to the conclusion that tricyclic antidepressants are more effective agents in the treatment of endogenous depression or major depression with melancholic features. Figure 1 characterizes the response rates generated by the melancholia efficacy studies to date. Despite the wide use of SSRIs as well as venlafaxine and nefazodone in the treatment of depression, it seems reasonable that clinicians subtype their depressed patients and treat melancholic patients first with a course of tricyclic antidepressants. Logically, it would seem that nortriptyline would be the TCA of choice over clomipramine or amitriptyline owing to its lack of adverse effects relative to the other TCAs and its ease of dosing (Roose et al., 1994). Initiating nortriptyline treatment at 1 mg/kg/d and adjusting the dose within the therapeutic range of 50-150 ng/ml seems to be the most straightforward dosing approach to treatment. For melancholic patients who have not responded to a SSRI, several options can be considered. These include initially a TCA trial and if this fails a venlafaxine trial or ECT.
Figure 1. Percentage of partial response (>50% HAMD decrease or HAMD <16) of full response (HAMD <7, 8, or 10) rates for TCAs, SSRIs, venlafaxine, and nefazodone in endogenous depression or major depression with melancholic features.
References
American Psychiatric Association (1980) DSM-III Diagnostic and Statistical Manual of Mental Disorders. 3rd ed. APA, Washington, DC.
Beasley, C.M., Holman, S.L., and Potvin, J.H. (1993) Fluoxetine compared with imipramine in the treatment of inpatient depression. Ann. Clin. Psychiatry 5,199-207.
Benkert O, Grunder G, Wetzel H, Hackett D (1996) A randomized, couble-blind comparison of a rapidly escalating dose of venlafaxine and imipramine in inpatients with major depression and melancholia. J Psychiatr Res 30,441-51.
Bowden, C.L., Schatzberg, A.F., Rosenbaum, A., Contreras, S.A., Samson J.A., Dessain E., and Sayler, M. (1993) Fluoxetine and desipramine in major depressive disorder. J. Clin. Psychopharmacol. 13,305-310.
Clerc GE, Ruimy P, Verdeau-Pailles J (1994) A double-blind comparison of venlafaxine and fluoxetine in patients hospitalized for major depression and melancholia. Int Clin Psychopharmacol 9,139-143.
Danish University Antidepressant Group (1986) Citalopram: clinical effect profile in comparison with clomipramine: a controlled multicenter study. Psychopharmacology (Berl) 90,131-138.
Danish University Antidepressant Group (1990) Paroxetine: a selective serotonin reuptake inhibitor showing better tolerance, but weaker antidepressant effect than clomipramine in a controlled multicenter study. J. Affect. Disord. 18,289-299.
Guelfi JD, White C, Hackett D et al (1995). Effectiveness of venlafzine in patients hosptialized for major depression and melancholia. J Clin Psychiatry 56,450-458.
Heiligenstein, J.H., Tollefson, G.D., and Faries, D.E. (1994) Response patterns of depressed outpatients with and without melancholia: a double-blind, placebo-controlled trial of fluoxetine versus placebo. J. Affect. Disord. 30,163-173.
Janakiramaiah N, Motreja S, Gangadhar BN et al (1998). Once versus three times weekly ECT in melancholia: a randomized controlled trial. Acta Psychiatr Scand 98:316-20.
Laakman, G., Blaschke, D., Engel R., and Schwarz, A. (1988) Fluoxetine vs amitriptyline in the treatment of depressed out-patients. Br. J. Psychiatry 153(suppl 3),64-68.
Marcus R, Mendels J (1996). Nefazodone in the treatment of severe, melancholic and recurrent depression. J Clin Psychiatry 57[suppl2]:19-23.
Mendels J, Kiev A, Fabre LF (1999). Double-blind comparison of citalopram and placebo in depressed outpatients with melancholia. Depression Anxiety 9:54-60.
Mulsant BH, Pollock BG, Nebes RD, et al (1999). A double-blind randomized comparison of nortriptyline and paroxetine in the treatment of late-life depression: 6-week outcome. J Clin Psychiatry 60(suppl20):16-20.
Nelson, D.R., Thomas, D.R., and Johnson, A.M. (1989) Pharmacological effects of paroxetine after repeated administration to animals. Acta Psychiatrica Scand. 80:(suppl 35),21-23.
Nelson, J.C., Mazure, C.M., Bowers, M.B., and Jatlow, P.I. (1991) A preliminary, open study of the combination of fluoxetine and desipramine for rapid treatment of depression. Arch. Gen. Psychiatry 48:303-307.
Roose, SP, Glassman, AH, Attia, E, Woodring, S. (1994) Comparative efficacy of selective serotonin reuptake inhibitors and tricyclics in the treatment of melancholia. Am. J. Psychiatry 151,1735-1739.
Spitzer, R.L., Endicott, J. and Robins, E. (1978) Research Diagnostic Criteria. New York State Psychiatric Institute, New York.
Stage KB, Bech P, gram LF et al (1998). Are in-patient depressives more often of the melancholic subtype? Acta Psychiatr Scand 98:432-436.
Stuppaeck, C.H., Geretsegger, C., Whitworth, A.B., Schubert, H., Platz, T., Konig, P., Hinterhuber, H., and Felischhacer, W.W. (1994) A multicenter double-blind trial of paroxetine versus amitriptyline in depressed inpatients. J. Clin. Psychopharmacol. 14,241-246.
Thase ME, Entsuah AR, Rudolph RL (2001). Remission rates during treatment with venlafaxine or selective serotonin reuptake inhibitors. Br J Psychiatry 178:234-241.
Tignol, J., Stoker, M.J., Dunbar, G.C. (1992) Paroxetine in the treatment of melancholia and severe depression. Int. Clin. Psychopharmacol. 7,91-94.
Tollefson GD, Holman SL (1993) Analysis of the Hamilton Depression Rating Scale factors from a double-blind, placebo-controlled trial of fluoxetine in geriatric major depression. Int Clin Psychopharmacol 8:253-9.
Tzanakaki M, Guazzelli, Nimatoudis I, et al (2000). Increased remission rates with venlafaxine compared with fluoxetine in hsoptialixed paitents with major depression and melacholia. Int Clin Psychopharmacol 15:29-34.
Winokur G (1991) Mania and Depression, A Classification of Syndrome and Disease. Baltimore: Johns Hopkins University Press.
Zimmerman M, Stangl D, Pfohl B, and Coryell W. (1984) Diagnosing RDC endogenous depression and DSM-III melancholia [letter]. Psychiatry Res. 12,173-6.