Time and depression treatment: the value of early treatment response

Hans-Jürgen MÖLLER, MD, Prof
Michael RIEDEL, MD, Prof
Department of Psychiatry
Ludwig-Maximilians University

Time and depression
treatment: the value of
early treatment response

by H. J . Möl ler, F. H. Seemül ler, and M. Riedel ,Germany

Major depression is still today a devastating illness, currently reflected by the most recent World Health Organization statistic showing depression to be the leading cause of years lost due to disability. Although recent psychopharmacologic developments have resulted in progress, individualized psychopharmacology is in its early stages. One way to individualize treatment is to use predictors such as early response. Today, there is broad consensus among the research community that (i) early changes resulting from treatment of major depression occur within the first 2 weeks of antidepressant treatment; and (ii) they are highly predictive of the later outcome. Nevertheless, this fact is still largely ignored by most treatment guidelines in major depression. This review summarizes current knowledge of the methodological pitfalls in the assessment of early treatment response, discusses actual concepts of possible biological mechanisms involved in early drug response, and summarizes the current knowledge base regarding the potential of today’s available drugs to induce early treatment changes. Moreover, the reasons still contributing to the outdated belief of the delayed onset hypothesis, and the clinical implications, will be discussed.

Medicographia. 2010;32:139-145 (see French abstract on page 145)

Major depression is a severe, often recurrent, and ultimately disabling disorder. In Europe, about 33.3 million people suffer from unipolar depression. According to the latest World Health Organization statistic (2004), depression is the third most common medical condition causing moderate to severe disability in the world (Table I, page 140).1 Years lost due to disability measure the equivalent years of healthy life lost through time spent in states of less than full health. In this category, unipolar depression is the leading cause for years lost due to disability worldwide (Table II, page 140).

Table I
Table I.
Estimated prevalence
of moderate and severe
disability (millions)
for leading disabling
conditions by age, for
high-income and lowand
countries, 2004.

After reference 1: World
Health Organization Web
site. http://www.who.int/

Accessed November 30,
2009. Copyright © 2009,
World Health Organization.

Table II
Table II. Leading global causes of Years Lost to Disabilty (YLD) in high-income and low- and middle-income countries, 2004.

COPD, chronic obstructive pulmonary disease.
After reference 1: World Health Organization Web site. http://www.who.int/ healthinfo/global_burden_disease/2004_report_update/en/index.html. Accessed November
30, 2009. Copyright © 2009, World Health Organization.

One of the main problems in the treatment of major depression with antidepressant compounds lies in the difficulty of there being a lack of reliable predictors of antidepressant response or remission at the level of the individual. Obviously, a variety of components contribute to antidepressant response and remission, which results in difficulties in disentangling the determinants. Various clinical features involved have been proposed.2-4 Against this background of still not knowing which is “the right drug for the right person” from the beginning of treatment, the still widely-accepted delayed onset hypothesis bears some noticeable risks. In the worst case scenario, this hypothesis may in fact lead to initiation of rather unspecific trial ther- apy with an antidepressant compound (without considering subtype and individual psychopathology). After some waiting, and no visible changes after 1 or 2 months, a switch to another medication would follow. In such a “worst case scenario,” this very same procedure might be repeated until the patient remitted spontaneously after another couple of months. The underlying reason for this is the general and not infrequent assumption today that available antidepressant compounds take a minimum of 4 to 6 weeks to unfold their antidepressant properties. This assumption can lead to a certain clinical attitude, whereby the patient’s early reports and complaints are not listened to cautiously enough by the treating physician, due to the fact that changes in psychopathology— especially improvements—are naturally not expected. Thus, extremely valuable information for both doctor and patient is not infrequently lost. This position is still largely reflected by most treatment guidelines in major depression, which advise that a treatment effect cannot be expected before 4 to 6 weeks. Such guidelines originally derived from large randomized placebo-controlled trial data that usually showed placebo verum differences from week 3 onward. Today, however, the majority of the research community no longer holds such beliefs about treatment effects, as early antidepressant effects are known to occur within days or the first weeks after application of antidepressants.

The purpose of this review is thus to explore the risk of slow treatment response on the one hand, and the value of early changes in the treatment course on the other. Additionally, methodological reasons will be highlighted for the discrepancy between the delayed onset of antidepressant action hypothesis and the modern early improvement approach. Possible biologic mechanisms behind early improvement will be explored, and finally, current evidence regarding the ability of different antidepressant compounds to induce early symptom improvement will be summarized.

Risks of slow response to antidepressant treatment

True delayed onset of antidepressant efficacy has been connected withmany primary depression–related risks, as patients remain symptomatic and functionally impaired during the initial treatment time, as well as in connection with secondary psychosocial issues. A prolonged time to alleviation of the acute illness burden not only prolongs the vocational disability, but also bears an increasing risk for chronification of the current illness episode. This notion finds support in studies showing that the length of the current untreated episode of illness may be the strongest predictor of overall short-term and long-term outcome in major depression.5 The possible underlying pathophysiology for this, which has repeatedly been discussed in relation to several mental disorders, is a direct neurotoxic effect of the current depressive episode. Data in support of this notion come from a very recent high-resolution functional magnetic resonance imaging study on 20 medication- naïve patients with a first episode ofmajor depression who were compared with 20 healthy controls and 20 subjects who had fully recovered after a first episode. Patients in an acute episode showed significant enlargement of both amygdalae, whereas there was no difference between healthy controls and recovered subjects. Furthermore, amygdala size correlated significantly with the depression severity.

Another major issue that seems to be connected with the slow response of antidepressant treatment is suicidality. In a study by Jick and coworkers, an increased risk of suicidal behavior was especially noticed during the first 9 days of treatment.6 The odds for suicidal behavior among patients first prescribed an antidepressant 1 to 9 days before their index date (ie, occurrence of suicidal ideation) were 4.07 (95% confidence interval [CI], 2.89-5.74) compared with patients who were first prescribed an antidepressant 90 days or more before their index date. This is in line with several studies on suicidal acts in inpatients and outpatients, showing the highest risk to be within the first week or month of commencement of antidepressant treatment.7 The early occurrence of suicidal ideation has constantly been connected with a mismatch between early symptom improvements like increased psychic and physical energy, and the more gradual resolution of depressed mood and hopelessness.8 One major confounder in this context is clearly the fact that there is a large overlap between predictors of poor treatment response and occurrence of suicidality. Persistent hopelessness during treatment might be one of the major psychopathological symptoms accounting for suicidal ideation during treatment, as shown by Maria Oquendoo for unipolar and bipolar depressed subjects.9 The overlap between predictors of response and suicidality and the finding that suicidal acts occur soon after starting antidepressant treatment suggest that achieving a more rapid and enhanced treatment response, targeting core depressive symptoms including hopelessness, could help to substantially reduce the incidence.10

A delayed onset in the effect of antidepressant treatment can also be associated with secondary psychosocial losses. It has been proven fairly well that depression limits quality of life, particularly through its impairment of the cognitive skills necessary for work, creating and maintaining relationships, being productive, and functioning in multiple domains.11,12 Beside from such issues, a prolonged time to antidepressant treatment response also increases the patient’s subjective experience of a lack of treatment efficacy, which may lead to frustration and damage to the patient doctor relationship, usually finally resulting in poor compliance rates.13 This latter issue is also reflected by the high dropout rates in placebo-controlled trials for nonresponding subjects.

The unmet need for fast improvement in treating major depression: a new therapeutic paradigm?

Early improvement and onset of antidepressant action has been a matter of research and discussion for decades. Nowadays, there is strong evidence pointing in the direction of true early antidepressant effects. Since delayed onset of antide- pressant action bears considerable risks, would it not seem wise to further explore the potential role of early onset of antidepressant action as a surrogate end point for long-term sustained stability, given that at the same time it is associated with long-lasting benefits such as limitation of harmful neurobiological effects, limitation of poor outcome secondary to repeated depressive episodes, and limitation of enduring depressive symptoms? In a next step, specific therapeutic interventions leading to fast improvement could be developed.

But before the paradigm can be changed, one must think about the methodological problems and pitfalls that need to be considered in examining fast treatment response/effects. So far, the vastmajority of data on the topic, presented in summary below, have relied on post hoc analysis or meta-analysis of large-scale placebo-controlled or naturalistic studies. There are almost no prospective trials in this field.

Data from randomized controlled trials showing drug placebo differences not earlier than week 3 or 4 usually rely on significant verum placebo differences in mean scores on a rating scale. But clearly, such an approach cannot detect early significant symptom changes in individuals, which clinicians regularly observe. An advantage would be the incorporation of responder and remitter analysis of such data, which would allow consideration of significant treatment benefits at the individual level. But we would still need to keep in mind that the current cut-off point for a response, ie, a 50% improvement from baseline, might be too strict to detect slight but significant early changes. Thus a growing body of literature suggests that a cut-off of a 20% improvement from baseline on a depression rating scale might be sensitive enough to detect early changes.14

Next, we should reconsider the instruments currently used as the gold standard in depression research. The Hamilton Rating Scale for Depression (HAM-D), for example, is well known to not be very sensitive at detecting treatment changes, as opposed to the Montgomery–Asberg Depression Rating Scale, which was specifically developed for this.15 But there are also other aspects of the study design that need to be taken into account. Most trials of antidepressants use weekly or biweekly measurements, which are too infrequent and wide-apart to detect early changes that can occur within hours or days. A very useful tool may be online life-charting, which is nowadays freely available on several Web sites (eg, www.moods wings.net.au).

One last aspect concerns statistical analysis of early improvement, which bears some unique features; in this area, patients have most often been investigated at different time points that depend on the individual. Statistical analysis should ideally rely on mixed models, including varied assessments as a random variable. In a direct comparison with placebo or an active comparator, it may be most sensitive and appropriate to use a survival analytic approach, although othermethodologies can also provide useful information.16 Keeping all these limitations in mind, we will now briefly review the available evidence regarding the predictive ability of early response during antidepressant treatment.

Figure. Final remission rates of early improvers (20% reduction
on the Hamilton Rating Scale for Depression [HAM-D17] after
week 2) in a study of naturalistically-treated depressed inpatients.

After reference 14: Henkel V, Seemuller F, Obermeier M, et al. J Affect Disord. 2008;115:439-449. Copyright © 2008, Elsevier Inc.

How does early improvement predict later stable response and remission?

Traditionally it has long been thought that standard antidepressants take about 1 month for their antidepressant action to fully unfold, with a delayed onsetof actionof at least 2 weeks. Originally, Quitkin proposed in his pattern analytic approach that drug effects could not be observed before 3 weeks of treatment.17 An earlier improvement was supposed to be a placebo response, with a subsequent lack of sustained improvement, whereas the opposite was true for true drug responders who showed a delayed but sustained onset of response.17,18 More recently, this view has been questioned by a large number of authors who have not only emphasized that an earlier onset of response before 2 weeks is highly prevalent, but have also shown that it was highly predictive of later outcome.19-21 An early improvement was thus defined as a 20% reduction in the initial HAM-D17 score within the first 2 weeks. Henkel and co-workers recently demonstrated in a large naturalistically treated sample of 1014 depressed inpatients that about 80% of all early improvers achieved full response at the final end point, whereas only 50%of the nonearly improvers did so.14 Concerning remission, about 58%of all early improvers were also remitters at the final end point; by contrast, 63% of the non-improvers also became nonremitters at discharge (Figure).14 It appears that early improvement defined as a 20% reduction during the first 2 weeks may be an excellent trait variable for use in treatment decisions. As psychiatrists usually want to fully utilize the potential of each single drug and to minimize the risk of switching too early or changing a medication that might still start to work later on, the rate at which non-early improvers (corresponding to the sensitivity of early improvement as a predictor of later response) finally respond or remit might be the most clinically meaningful variable. In other words, if most nonearly responders stay nonremitters or nonresponders, then further and earlier therapeutic interventions are indicated in order to improve the outcome.

In line with the data of Henkel et al, but even more striking, are very recent pooled data from a meta-analysis of randomized controlled phase 3 trials involving 6562 patients carried out by the group of Armin Szegedi.22 The data showed that only 4% and 11% of the non-early improvers became stable remitters and responders, respectively, after 4 or more weeks. In other words, if a patient does not show at least some minimal improvement within the first 2 weeks, there is a 96% chance that they will also be a nonremitter after 4 or more weeks, and an 89% chance that they will be a nonresponder, if no changes are made to the medication regime.22 This leads to the conclusion that if after 2 weeks one can observe no improvement at all under a new antidepressant regimen, then the pharmacologic regimen should be adjusted or changed immediately rather than waiting for another 2 or 3 weeks. Before we go on deeper into the current knowledge of antidepressant compounds and their potential for inducing early symptomchanges, we will brieflydiscuss the underlying biological mechanisms possibly involved in early treatment changes.

Biologic mechanisms of early onset of antidepressant compounds

Due to methodological difficulties, knowledge of the biology of antidepressant action is still very sparse. A very recent review by Marchedo-Viera gives an excellent overview of this topic.8 Concerning the traditional view of a delayed onset, one of the most widespread theories is the two period model, initially proposed by Hyman and Nestler in 1996.23 In the first “initial phase,” there is a correction of presumably disturbed monoaminergic neurotransmission, which is followed by a second “adaptation phase,” during which there are enduring modulatory changes in critical cortical circuits related to long-term antidepressant response.

Today, a growing body of evidence supports the notion that mood disorders might develop from abnormalities in cellular neuronal plasticity cascades. The term“neuroplasticity,” which is regularly used in this context, refers to remodeling and development of new synapses and axonal and dendritic architecture, and the growth of new neurons. Amongst other neurotrophins and cytokines, one central factor that is involved in the regulation of neuroplasticity is brain-derived neurotrophic factor (BDNF). BDNF levels have also very constantly been associated with antidepressant response. In animal models, for example, bilateral infusion of BDNF in rodents has been shown to induce a fairly rapid antidepressant effect within 3 days after a single administration. This effect lasted for at least 10 days, which supports there being some degree of persistence.8,24

More recently, glutamate has been found to be a central agent involved in the modulation of neuroplasticity. For example, Zarate and coworkers were able to show that a single i.v. dose of ketamine, an N-methyl-D-aspartic acid (NMDA) receptor antagonist, could induce a rapid and sustained antidepressant effect (within 110 minutes; over a period of 2 weeks) in subjects with treatment-resistant depression, compared with placebo. Of the 17 subjects investigated, 1 day after the i.v. dose, 71% met response criteria and 29% met remission criteria, and 35% remained responders for 1 week.25 Based on that observation, Machado-Vieira and coworkers hypothesized that this effect may be the result of two processes.8,26 The quick initial resolution of the depressive core symptoms might bemore a result of an increase in glutamatergic throughput rather than a result of neuroplastic changes. Second, the sustained effect might be the result of early changes in neuroplasticity.

With regard to nonpharmacological interventions, sleep deprivation has also been proven to show early antidepressant action after the first night. Further developments like sleep deprivation combined with consecutive sleep phase advance,27 or combinations of sleep deprivation with lithium or antidepressant treatments,28 are known to produce sustained antidepressant effects. The rapid and early antidepressant effects of sleep deprivation have been ascribed to elevations in BDNF levels29 and changes in the glutamatergic throughput within the dorsolateral prefrontal cortex.30

The next paragraph summarizes the current knowledge of antidepressant compounds so far to have demonstrated early symptomatic improvement.

Onset of response in available antidepressants

The demonstration of an early onset of response goes far back to the 80s, when Katz and coworkers began challenging the results of the landmark study of Quitkin showing that early improvement was associated with an assumed placebo response, as opposed to the later and sustained onset of antidepressant treatment action. Katz originally reported that an onset of action occurred within the first 10 days of treatment with tricyclic antidepressants. Unfortunately, this trial did not include a placebo control group.31 A little later, these shortcomings were overcome, and treatment-specific early changes were demonstrated in a study comparing desipramine and paroxetine with placebo. In this trial, significant early treatment effects occurred in the verum group, but not in the placebo group. Moreover, these effects were highly predictive of lat- er response. Subsequently, Nierenberg demonstrated that more then 50% of patients ultimately responding to fluoxetine showed early symptom improvements after week 2. The very same group also showed that non-early improvement was highly predictive of poor 8-week outcome.19

In 2003, Szegedi and colleagues studied early improvement in a randomized controlled trial comparing mirtazapine and paroxetine in patients with major depressive episode. Early improvement as measured with a 20% reduction in the HAM-D total score was present as soon as week 2, and was highly predictive of later stable response and remission for both drugs. Less than 10% of patients not improving after week 2 became responders or remitters thereafter. Most recently, Szegedi replicated his findings in the previously described meta-analysis, using a dataset from randomized active or placebo-controlled trials of mirtazapine, in 6907 inpatients and outpatients. The classes of active comparators included a serotonin norepinephrine reuptake inhibitor (SNRI; venlafaxine), SSRIs (paroxetine, fluoxetine, citalopram, sertraline, and fluvoxamine), tricyclics (amitriptyline, doxepin, and climipramine), as well as the tetracyclics maprotiline and trazodone. A total of 52% were taking mirtazapine. The majority of patients had at least a 20% reduction in the HAM-D17 total score by the end of week 2. Across all treatments, early improvement was a highly sensitive predictor of stable response (81%- 98%) and stable remission (87%-100%), although specificity was limited (43%-60% and 30%-53% for response and remission, respectively).22

Three years earlier, a meta-analysis focusing on SSRIs revealed very similar results. A total of 50 randomized controlled trials on fluoxetine, fluvoxamine, citalopram, escitalopram, sertraline or paroxetine including 5872 patients were analyzed.32 The main results revealed that treatment with SSRIs rather than with placebo was associated with clinical improvement by the end of week 1. In addition, the chance of achieving full response after week 1 (50% improvement on HAM-D) was significantly higher with an SSRI than placebo.

Most recent analyses of the first double-blind placebo-controlled trial on agomelatine including an active comparator (paroxetine) give additional support for an early onset of antidepressant action. The analysis for the time to first response indicated an initial significant effect (superiority over placebo) after 2 weeks for 25 mg/day of agomelatine. By contrast, the first significant advantages with paroxetine were only seen after 4 weeks.33,34 Agomelatine has also shown earlier signs of improvement compared with venlafaxine on the Clinical Global Impression scale after 1 week of treatment.35 With regard to the effect of combination or augmentation strategies, very little data is currently available. Henkel and coworkers tested the clinical utility of predicting treatment outcome in five different treatment subgroups (two subgroups with monotherapy [a compound from the two classes of new-generation antidepressant SNRIs or SSRIs] versus three subgroups with combination therapy). In the three combination therapy subgroups, an antidepressant compound along with a tranquilizer, antipsychotic compound, or mood stabilizer (lithium or an antiepileptic compound) was investigated. Sensitivity values were rather consistent across all treatment subgroups, ranging from 70% to 100% for both response and remission.14

All in all, every substance approved for the treatment of major depression appears to have inherited the potential to induce early treatment changes. Summarizing results of the previously cited trials, one possible important interpretation in line with the aforementioned biological and pharmacological mechanisms was recently proposed by Stassen and coworkers. His analysis used probably the most elaborate statistical method for the investigation of early treatment changes on an individual level. Here, a two-dimensional cure model disentangling the two crucial aspects of therapeutic response was applied, namely symptom improvement (incidence) and timing of improvement (latency), using a random-effect mixed model for a pooled dataset of 2848 patients treated with placebo or one of seven different antidepressants. In themain, there was no indication of any delayed onset of antidepressant drug response. Instead the authors found a highly individual time pattern of recovery, along with continuous distributions of the time spent to the onset of improvement across all treatments and placebo. Effective antidepressant treatment appeared to trigger and maintain conditions necessary for recovery. Thus, affectively ill patients might be likely to possess a common “resilience-like mechanism” that largely controls recovery and that is inducible by sufficient antidepressant treatment.36


In summary, these findings may lead to the conclusion that careful following of the very early treatment course of depressed patients under antidepressant therapy can give extremely helpful and valuable information for improving individualization of treatment. _


1. The Global Burden of Disease: 2004 Update. World Health Organization Web site. http://www.who.int/healthinfo/global_burden_disease/2004_report_update/ en/index.html. Accessed November 30, 2009.
2. Hirschfeld RM, Russell JM, Delgado PL, et al. Predictors of response to acute treatment of chronic and double depression with sertraline or imipramine. J Clin Psychiatry. 1998;59:669-675.
3. Kozel FA, Trivedi MH, Wisniewski SR, et al. Treatment outcomes for older depressed patients with earlier versus late onset of first depressive episode: a Sequenced Treatment Alternatives to Relieve Depression (STAR*D) report. Am J Geriatr Psychiatry. 2008;16:58-64.
4. Trivedi MH, Rush AJ, Wisniewski SR, et al. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. Am J Psychiatry. 2006;163:28-40.
5. Bagby RM, Ryder AG, Cristi C. Psychosocial and clinical predictors of response to pharmacotherapy for depression. J Psychiatry Neurosci. 2002;27:250-257.
6. Jick H, Kaye JA, Jick SS. Antidepressants and the risk of suicidal behaviors. JAMA. 2004;292:338-343.
7. Mortensen PB, Agerbo E, Erikson T, Qin P, Westergaard-Nielsen N. Psychiatric illness and risk factors for suicide in Denmark. Lancet. 2000;355:9-12.
8. Machado-Vieira R, Salvadore G, Luckenbaugh DA, Manji HK, Zarate CA, Jr. Rapid onset of antidepressant action: a new paradigm in the research and treatment of major depressive disorder. J Clin Psychiatry. 2008;69:946-958.
9. Oquendo MA, Bongiovi-Garcia ME, Galfalvy H, et al. Sex differences in clinical predictors of suicidal acts after major depression: a prospective study. Am J Psychiatry. 2007;164:134-141.
10. Brent DA, Emslie GJ, Clarke GN, et al. Predictors of spontaneous and systematically assessed suicidal adverse events in the treatment of SSRI-resistant depression in adolescents (TORDIA) study. Am J Psychiatry. 2009;166:418-426.
11. Wells KB, Stewart A, Hays RD, et al. The functioning and well-being of depressed patients. Results from the Medical Outcomes Study. JAMA. 1989;262:914-919.
12. Hirschfeld RM. The role of primary care in the treatment of depression. Am J Manag Care. 2000;6(2 Suppl):S59-S65.
13. Mendlewicz J. Sleep disturbances: core symptoms of major depressive disorder rather than associated or comorbid disorders. World J Biol Psychiatry. 2009; 10:269-275.
14. Henkel V, Seemuller F, Obermeier M, et al. Does early improvement triggered by antidepressants predict response/remission? Analysis of data from a naturalistic study on a large sample of inpatients with major depression. J Affect Disord. 2008;115:439-449.
15. Montgomery SA, Asberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry. 1979;134:382-389.
16. Montgomery SA, Bech P, Blier P, et al. Selecting methodologies for the evaluation of differences in time to response between antidepressants. J Clin Psychiatry. 2002;63:694-699.
17. Quitkin FM, Rabkin JD, Markowitz JM, Stewart JW, McGrath PJ, Harrison W. Use of pattern analysis to identify true drug response. A replication. Arch Gen Psychiatry. 1987;44:259-264.
18. Quitkin FM, McGrath PJ, Stewart JW, Taylor BP, Klein DF. Can the effects of antidepressants be observed in the first two weeks of treatment? Neuropsychopharmacology. 1996;15:390-394.
19. Nierenberg AA, McLean NE, Alpert JE,Worthington JJ, Rosenbaum JF, Fava M. Early nonresponse to fluoxetine as a predictor of poor 8-week outcome. Am J Psychiatry. 1995;152:1500-1503.
20. Stassen HH, Angst J, Delini-Stula A. Delayed onset of action of antidepressant drugs? Survey of results of Zurich meta-analyses. Pharmacopsychiatry. 1996; 29:87-96.
21. Szegedi A, Muller MJ, Anghelescu I, Klawe C, Kohnen R, Benkert O. Early improvement under mirtazapine and paroxetine predicts later stable response and remission with high sensitivity in patients with major depression. J Clin Psychiatry. 2003;64:413-420.
22. Szegedi A, JansenWT, vanWilligenburg AP, van derME, Stassen HH, ThaseME. Early improvement in the first 2 weeks as a predictor of treatment outcome in patients with major depressive disorder: a meta-analysis including 6562 patients. J Clin Psychiatry. 2009;70:344-353.
23. Hyman SE, Nestler EJ. Initiation and adaptation: a paradigm for understanding psychotropic drug action. Am J Psychiatry. 1996;153:151-162.
24. Shirayama Y, Chen AC, Nakagawa S, Russell DS, Duman RS. Brain-derived neurotrophic factor produces antidepressant effects in behavioral models of depression. J Neurosci. 2002;22:3251-3261.
25. Zarate CA Jr, Singh JB, Carlson PJ, et al. A randomized trial of an N-methyl-Daspartate antagonist in treatment-resistant major depression. Arch Gen Psychiatry. 2006;63:856-864.
26. Macedo-Soares MB, Moreno RA, Rigonatti SP, Lafer B. Efficacy of electroconvulsive therapy in treatment-resistant bipolar disorder: a case series. J ECT. 2005;21:31-34.
27. Berger M, Vollmann J, Hohagen F, et al. Sleep deprivation combined with consecutive sleep phase advance as a fast-acting therapy in depression: an open pilot trial in medicated and unmedicated patients. Am J Psychiatry. 1997;154: 870-872.
28. Leibenluft E, Moul DE, Schwartz PJ, Madden PA, Wehr TA. A clinical trial of sleep deprivation in combination with antidepressant medication. Psychiatry Res. 1993;46:213-227.
29. Gorgulu Y, Caliyurt O. Rapid antidepressant effects of sleep deprivation therapy correlates with serum BDNF changes in major depression. Brain Res Bull. 2009;80:158-162.
30. Murck H, Schubert MI, Schmid D, Schussler P, Steiger A, Auer DP. The glutamatergic system and its relation to the clinical effect of therapeutic sleep deprivation in depression—an MR spectroscopy study. J Psychiatr Res. 2009;43: 175-180.
31. Katz MM, Koslow SH, Maas JW, et al. The timing, specificity and clinical prediction of tricyclic drug effects in depression. Psychol Med. 1987;17:297-309.
32. Taylor MJ, Freemantle N, Geddes JR, Bhagwagar Z. Early onset of selective serotonin reuptake inhibitor antidepressant action: systematic review and metaanalysis. Arch Gen Psychiatry. 2006;63:1217-1223.
33. Eser D, Baghai TC, Moller HJ. Evidence of agomelatine’s antidepressant efficacy: the key points. Int Clin Psychopharmacol. 2007;22(suppl 2):S15-S19.
34. Loo H, Hale A, D’haenen H. Determination of the dose of agomelatine, a melatoninergic agonist and selective 5-HT(2C) antagonist, in the treatment of major depressive disorder: a placebo-controlled dose range study. Int Clin Psychopharmacol. 2002;17:239-247.
35. Lemoine P, Guilleminault C, Alvarez E. Improvement in subjective sleep in major depressive disorder with a novel antidepressant agomelatine: randomized double-blind comparison with venlafaxine. J Clin Psychiatry. 2007;68:1723-1732.
36. Stassen HH, Angst J, Hell D, Scharfetter C, Szegedi A. Is there a common resilience mechanism underlying antidepressant drug response? Evidence from 2848 patients. J Clin Psychiatry. 2007;68:1195-1205.

Keywords: depression; early response; predictor; treatment effect; remission