Psychopharmacology

, Volume 188, Issue 3, pp 273–280

Dose–response relationship of duloxetine in placebo-controlled clinical trials in patients with major depressive disorder

Authors

    • Psychiatric Research UnitFrederiksborg General Hospital
  • Daniel K. Kajdasz
    • Lilly Research LaboratoriesLilly Corporate Center
  • Vibeke Porsdal
    • Eli Lilly A/S
Original Investigation

DOI: 10.1007/s00213-006-0505-1

Cite this article as:
Bech, P., Kajdasz, D.K. & Porsdal, V. Psychopharmacology (2006) 188: 273. doi:10.1007/s00213-006-0505-1
  • 175 Views

Abstract

Rationale

The optimal dose for achieving the maximum antidepressive effect of selective serotonin reuptake inhibitors (SSRIs) or serotonin-noradrenalin reuptake inhibitors (SNRIs) remains a controversial issue. The varying sensitivity of scales that measure the severity of depression is one of the many factors affecting the evaluation of the dose–response relationship with antidepressants.

Objectives

To determine if the 6-item Hamilton rating scale for depression (HAM-D6) demonstrates a clearer association between dose and antidepressive effect compared with the 17-item Hamilton rating scale for depression (HAM-D17) for fixed doses of duloxetine hydrochloride (40, 60, 80, and 120 mg daily) from six double-blind, randomized, placebo-controlled clinical trials assessing safety and efficacy in the acute treatment of patients with DSM-IV-defined major depressive disorder (MDD).

Methods

Mantel–Haenszel adjusted effect sizes were determined by dose for change from baseline to endpoint in HAM-D6 and HAM-D17 scores from the six studies. To confirm, assessments were repeated on the subset of the population corresponding to the 70% of patients with the longest duration of treatment regardless of study, treatment, dose, geography, or completion status.

Results

For the majority of assessments, HAM-D6 effect sizes were numerically larger than those estimated from the HAM-D17. Findings support that duloxetine 60 mg daily is the best effective dose.

Conclusions

In this assessment of patients with MDD, the HAM-D6 was shown to be more sensitive compared with the HAM-D17 at detecting treatment effects. These findings are consistent with published results of other effective antidepressants.

Keywords

DuloxetineCymbaltaDepressionDose–responseHAM-D6HAM-D17

Introduction

The optimal dose for achieving the antidepressive effect of selective serotonin reuptake inhibitors (SSRIs) still remains a controversial issue, even in patients with unipolar, depressive disorders (Rifkin 1997). High placebo response and the questionable sensitivity of the depression rating scales such as the Hamilton Depression Scale (HAM-D) (Hamilton 1967) are factors that impact the evaluation of the dose–response relationship (Preskorn 1998). Fixed-dose, placebo-controlled trials with fluoxetine 5–60 mg daily administered over ≥3 weeks have shown that 20 mg daily is the optimal dose when adverse events are considered (Wernicke et al. 1987, 1988; Dunlop et al. 1990). Similarly, sertraline 50 mg daily appears to be the optimal dose when compared to fixed doses of 100 and 200 mg daily in a placebo-controlled trial (Fabre et al. 1995). However, although the difference in score change between sertraline and placebo was 3 points on the HAM-D, which is a clinically significant difference according to the National Institute for Health and Clinical Excellence (NICE) as emphasized by Moncrieff and Kirsch (2005), this result should be evaluated against the pooled standard deviation of the two treatments, i.e., the effect size. In the study by Fabre et al. (1995), the pooled standard deviation was 8 and the effect size consequently 3/8 or 0.38 (Hedges and Olkin 1985).

The optimal dose of the serotonin-noradrenalin reuptake inhibitors (SNRIs) is also controversial. A metaanalysis of the antidepressive efficacy of milnacipran (Lecrubier et al. 1996) showed that 75 mg daily was more effective than both 50 and 200 mg daily. The first fixed-dose, placebo-controlled trial using venlafaxine 25–200 mg daily (Mendels et al. 1993) in patients with major depression showed very small effect sizes after 6 weeks of therapy based on the HAM-D total score. However, in another placebo-controlled trial, Rudolph et al. (1998) showed that fixed doses of venlafaxine at 225 and 375 mg daily were significantly superior to 75 mg daily. Recently, Gex-Fabry et al. (2004) demonstrated a plasma concentration–response relationship when using a fixed dose of venlafaxine 300 mg daily, and also demonstrated that at this level the noradrenalin reuptake inhibition is operating (Gex-Fabry et al. 2004). In placebo-controlled trials with venlafaxine vs the SSRIs, Entsuah et al. (2002) identified higher effect sizes on the different subscales derived from HAM-D than on the full HAM-D17, but they included no results on dose–response or plasma-levels effect relationship. Among these HAM-D subscales, e.g., the HAM-D6 (Bech et al. 1975, 1984; O’Sullivan et al. 1997), the Maier–Philipp subscale (Maier and Philipp 1985), and the Gibbons subscale (Gibbons et al. 1993), the HAM-D6 is the one that has been used most systematically in both dose–response trials and in plasma–effect trials [e.g., Reisby et al. (1979), DUAG (1999), Bech et al. (2002, 2004)].

Recent psychometric evaluations of the HAM-D17 (Bagby et al. 2004; Licht et al. 2005) have questioned the validity of this version as an outcome measure of the antidepressive effect and have found the HAM-D6 valid in this respect. In terms of interrater reliability the HAM-D6 has been found superior to the Clinical Global Impression ratings (Ruhe et al. 2005).

Duloxetine hydrochloride (Cymbalta®) is an SNRI with a more balanced reuptake inhibition of serotonin and noradrenalin than venlafaxine (Bymaster et al. 2001). It has been shown that a fixed dose of duloxetine 60 mg daily is an effective dose when compared to placebo in patients with major depression (Detke et al. 2002a,b). Duloxetine is currently approved for the treatment of major depressive disorder (MDD) and the management of diabetic peripheral neuropathic pain in the US, European Union, and numerous other countries.

When profiling the dose–response relationship or the plasma concentration–effect relationship of clomipramine, only the HAM-D6, (Bech et al. 2002; Bagby et al. 2004) was valid, while the HAM-D total was not (Reisby et al. 1979; DUAG 1999). To validate this finding between the HAM-D6 and HAM-D17, we evaluated the dose–response relationship of duloxetine in all placebo-controlled, fixed-dose registration trials that were conducted as part of the clinical development plan for duloxetine in patients with MDD. These studies utilized fixed doses of duloxetine ranging from 40 mg daily (20 mg bid) to 120 mg daily (60 mg bid) using the change from baseline in the HAM-D17 total score as the a priori primary outcome measure. Four of these studies included a 20-mg daily paroxetine arm as an active control. Using effect size analysis, we aim to demonstrate a more clear association between dose and antidepressive effect based on the HAM-D6 compared with the HAM-D17 for duloxetine 40–120 mg daily in the acute treatment of patients with MDD.

Materials and methods

The selection of studies included in this dose–response analysis of duloxetine was based on the approach by Peto (1987), Bulpitt (1988), and Schmid et al. (1991), suggesting that trials should be included independent of results, the selected trials should not be biased, (i.e., only properly randomized and blinded placebo-controlled trials are to be considered), and the trials should address the same therapeutic issue in a well-defined patient population using similar design and outcome variables. Six randomized, double-blind, placebo-controlled, short-term efficacy studies of duloxetine 40–120 mg daily were sufficiently similar in regard to study design, inclusion/exclusion criteria, and outcome variables. All studies were performed in full accordance with the Declaration of Helsinki and its later revisions.

In all trials, the participating patients were male or female outpatients, ≥18 years of age, who met the DSM-IV criteria for a major depressive episode (single or recurrent). Patients had a HAM-D17 score ≥15 and a Clinical Global Impression of Severity (CGI-S) score ≥4 at the screening and pretreatment study visits. Patients were excluded from the trials if they had any current primary DSM-IV Axis I diagnosis other than MDD, a previous diagnosis of psychosis, bipolar disorder or schizoaffective disorder, an anxiety disorder as a primary diagnosis within the past year, or an Axis II disorder which could interfere with protocol compliance. Patients were also excluded from participation if they had lack of response during their current episode of depression to two or more adequate courses of antidepressant therapy, were judged to be at serious suicidal risk, had a serious medical illness, or had a history of substance abuse or dependence within the past year. In general, the use of concomitant medications with primarily central nervous system activity was not allowed in these studies. Patients were allowed the episodic use of benzodiazepines or certain hypnotics during study participation for no more than 6 days total.

The duration of the acute-treatment phase of the studies was 8 or 9 weeks depending on the specific protocol. The patients were evaluated weekly during the first weeks and then biweekly. In four studies, there were HAM-D17 data from 2, 4, 6, and 8 weeks on treatment and in the other two studies, there were HAM-D17 data from 2, 5, 7, and 9 weeks on treatment. Data were pooled after 2, 4–5, 6–7, and 8–9 weeks, respectively. In two of the studies, patients who responded to therapy in the acute phase could enter a 28-week continuation phase. Data from the continuation phase are not used in the analysis presented in this study.

Table 1 shows the characteristics of the six duloxetine trials. The studies were conducted as twin trials under three protocols. Studies 1 and 2 were identical US-based trials with four arms each: duloxetine 40 mg daily (20 mg bid), duloxetine 80 mg daily (40 mg bid), paroxetine 20 mg daily (20 mg qd), and placebo. Studies 3 and 4 were European-based trials with four arms each: duloxetine 80 mg daily (40 mg bid), duloxetine 120 mg daily (60 mg bid), paroxetine 20 mg daily (20 mg qd), and placebo. Studies 5 and 6 were US-based trials with two arms each: duloxetine 60 mg (60 mg qd) and placebo. Because geography variability (i.e., US vs EU trials) was not present within each study, we have performed a post hoc comparison of the baseline scores of HAM-D17 and HAM-D6 in accordance with Niklson and Reimitz (2001).
Table 1

Summary of six double-blind, fixed-dose, placebo-controlled, randomized, multi-center registration studies assessing the safety and efficacy of duloxetine hydrochloride for the treatment of major depressive disorder

Study

Studies 1 and 2

Studies 3 and 4

Studies 5 and 6

Country

USA

Europe (7 countries)

USA

Arms

Duloxetine 40 mg/day

Duloxetine 80 mg/day

Duloxetine 60 mg/day

Duloxetine 80 mg/day

Duloxetine 120 mg/day

Paroxetine

Paroxetine

Placebo

Placebo

Placebo

Duration of acute treatment

8 weeks

8 weeks

9 weeks

Number of patients, randomizeda

40 mg: 91

40 mg: 86

80 mg: 95

80 mg: 93

60 mg: 123

60 mg: 128

80 mg: 84

80 mg: 91

120 mg: 93

120 mg: 103

Placebo: 90

Placebo: 89

Placebo: 93

Placebo: 99

Placebo: 122

Placebo: 139

Number of patients, completed (acute phase)a

40 mg: 61

40 mg: 55

80 mg: 83

80 mg: 83

60 mg: 76

60 mg: 70

80 mg: 56

80 mg: 53

120 mg: 84

120 mg: 90

Placebo: 61

Placebo: 52

Placebo: 75

Placebo: 90

Placebo: 84

Placebo: 87

Mean age, years

44

41

43

45

42

41

Sex, % women

62

62

73

70

67

69

HAM-D17 at baseline, mean

17.7

17.9

20.1

21.1

21.3

20.4

Authorb

Nemeroff et al. 2002

Goldstein et al. 2004

Detke et al. 2004

Accepted for publication in European Psychiatry

Detke et al. 2002a

Detke et al. 2002b

aIn duloxetine and placebo groups

bResults of all studies are also available at http://www.lillytrials.com

Statistical analysis

Outcome measure

The 6-item subscale of the HAM-D, the HAM-D6, that includes the depression-specific items was used in this analysis (Bech et al. 1975, 1981; Bagby et al. 2004; Licht et al. 2005). The HAM-D6 score is the sum of the responses to the six items pertaining to depressed mood, guilt feelings, work and interests, psychomotor retardation, psychic anxiety, and general somatic symptoms. The HAM-D6 total score has been shown to be a more valid measure of depression severity than the HAM-D17 because of its focus on the core symptoms of depression (Bagby et al. 2004;Bech 2002; Licht et al. 2005; Ruhe et al. 2005).

Analyses of dose–response relationship

Effect size for a scale was defined as the difference in mean change from baseline to the respective timepoint in HAM-D scores between patients treated with duloxetine and placebo divided by the pooled standard deviation for the two treatment groups (Hedges and Olkin 1985). Baseline was defined as the measure collected at or most recently before randomization while endpoint was defined as the last nonmissing observation obtained from the patient during the acute treatment phase of the studies. All patients with ≥1 postbaseline HAM-D measure were included in the assessments, and effect size assessments based on weeks on treatment included all subjects with HAM-D observations at the corresponding study visit. To check the consistency of the findings, effect sizes by dose were calculated for change from baseline to endpoint for the subset of patients corresponding to the 70% of patients who maintained the longest duration of treatment without regard to study, geography, treatment group, dose, or completion status (Angst et al. 1989).

The overall effect size for each dose was determined from individual studies using the Mantel–Haenszel metaanalytic approach. Study-specific effect sizes were weighted by their inverse variance for each dose (Engels et al. 2000). Ninety-five percent confidence intervals (95% CI) were calculated by applying the normal approximation method using the standard error of the overall effect size estimate for each dose. It has been shown that an effect size of approximately 0.40 is the minimum effective response both in trials with SSRIs (Bech et al. 2000, 2002; Faries et al. 2000) and dual-acting antidepressants such as mirtazapine (Bech 2001) when using HAM-D6 as the outcome measure to identify clinically significant levels of response.

Results

Table 2 shows the demographics and baseline disease severity by treatment group. There were no differences in gender, origin, age, or baseline HAM-D17 scores between the 632 patients receiving placebo and the 987 patients receiving duloxetine. The majority of subjects were Caucasian females averaging in age in their early forties. Mean HAM-D17 scores at baseline were 19.7 in the placebo group and 19.8 in the duloxetine group. Rates of discontinuation due to adverse events were 11.9% for the 40 mg daily group, 13.1% for the 60 mg daily group, 9.6% for the 80 mg daily group, and 3.6% for the 120 mg daily group. The overall rate of discontinuation due to adverse events for the placebo patients was 4.0%. Study-specific, placebo-referenced relative risks (with 95% CIs) were 1.9 (1.0–3.9), 3.8 (1.9–7.8), 2.2 (1.3–4.0), and 1.4 (0.4–4.2), respectively.
Table 2

Patient demographic and disease severity characteristics at baseline for all randomized patients

Characteristic

Placebo (N=632)

Duloxetine (N=987)

P valuea

Gender

 Female, %

432 (68.4)

661 (67.0)

0.571

 Male, %

200 (31.6)

326 (33.0)

Ethnicity

 African descent, %

36 (5.7)

46 (4.7)

0.114

 Caucasian, %

549 (86.9)

882 (89.4)

 Hispanic, %

43 (6.8)

43 (4.4)

 Other, %

4 (0.6)

16 (1.6)

Age, years, mean (SD)

42.4 (13)

42.9 (12.5)

0.689

HAM-D total, mean (SD)

19.7 (4.2)

19.8 (4.5)

0.130

CGI-S, mean (SD)

4.2 (0.6)

4.2 (0.6)

0.995

Categorical outcomes were assessed using the CMH test for general association controlling for study. Continuous outcomes were assessed using an ANOVA model containing the mean effects for treatment group and study.

CGI-S Clinical global impressions of severity, SD standard deviation

aP values for between treatment group differences.

Table 3 shows the effect size for the HAM-D6 and HAM-D17 after 2, 4–5, 6–7, and 8–9 weeks of therapy and at endpoint using last observation carried forward (LOCF). After 2 weeks of therapy, duloxetine 60 mg daily, on the HAM-D6 obtained an effect size >0.40 that was sustained until endpoint. Duloxetine 40 mg daily never reached an effect size of 0.40, while both duloxetine 80 and 120 mg obtained levels >0.40 after 6 weeks of therapy. Numerically, the effect sizes for duloxetine 120 mg daily were higher than those for 60 or 80 mg daily, but the differences were not statistically significant based on the overlap of 95% confidence intervals.
Table 3

Effect sizes by daily dose based on the change from baseline in HAM-D6 and HAM-D17 scores

Weeks of therapya

Effect sizes (95% CI) duloxetine dose

40 mg daily

60 mg daily

80 mg daily

120 mg daily

HAM-D6

HAM-D17

HAM-D6

HAM-D17

HAM-D6

HAM-D17

HAM-D6

HAM-D17

Week 2

0.16 (−0.05, 0.38)

0.18 (−0.04, 0.40)

0.41 (0.22, 0.59)

0.33 (0.14, 0.51)

0.13 (−0.02, 0.28)

0.13 (−0.02, 0.28)

0.06 (−0.14, 0.26)

0.09 (−0.11, 0.30)

Week 4/5

0.28 (0.05, 0.50)

0.28 (0.05, 0.50)

0.46 (0.26, 0.66)

0.35 (0.15, 0.54)

0.28 (0.12, 0.43)

0.18 (0.02, 0.33)

0.37 (0.17, 0.58)

0.35 (0.15, 0.56)

Week 6/7

0.38 (0.13, 0.62)

0.34 (0.09, 0.58)

0.54 (0.33, 0.74)

0.47 (0.27, 0.68)

0.46 (0.29, 0.62)

0.34 (0.18, 0.50)

0.58 (0.37, 0.79)

0.50 (0.29, 0.71)

Week 8/9

0.31 (0.06, 0.57)

0.34 (0.08, 0.60)

0.51 (0.29, 0.72)

0.47 (0.26, 0.69)

0.48 (0.31, 0.65)

0.37 (0.21, 0.54)

0.61 (0.40, 0.83)

0.53 (0.31, 0.75)

LOCF endpoint

0.30 (0.08, 0.52)

0.30 (0.09, 0.52)

0.51 (0.32, 0.69)

0.46 (0.28, 0.65)

0.49 (0.34, 0.64)

0.41 (0.26, 0.56)

0.57 (0.37, 0.78)

0.49 (0.28, 0.69)

aPatients in the studies using duloxetine 60 mg daily were assessed at weeks 2, 5, 7, and 9; in the other studies patients were assessed at weeks 2, 4, 6, and 8.

The effect size pattern by dose was less clear for the HAM-D17. Duloxetine 60 mg daily obtained an effect size >0.40 after 7 weeks of therapy while duloxetine 80 mg daily demonstrated at effect size ≥0.40 only at the LOCF endpoint assessment. Numerically, duloxetine 60 mg daily and 120 mg daily showed similar effect sizes.

As what appears in Table 1, 505 or 87.7% of the 576 patients included in the EU trials (studies 3 and 4) completed the planned treatment of 8 weeks, while 655 or 62.8% of the 1,043 patients included in the US trials completed (studies 1, 2, 5, and 6). Based on these findings, the subset of patients who were in the 70% percentile with regard to the longest duration of treatment was considered according to Angst et al. (1989). Tables 4, 5, and 6 show the baseline HAM-D17 and HAM-D6 scores for all included patients (Table 4), for the subset of patients with the longest duration of treatment (Table 5) and for those patients within the subset defined in Table 5 who have a HAM-D17 total score of 18 or more at baseline (Table 6). The results showed (Table 4) that HAM-D17 was scored highest in the US trials and HAM-D6 highest in the EU trials. Due to the large number of observations, these differences were statistically significant although the numerical difference was small and not of great clinical significance. The same pattern was found for the subset of patients with the longest duration of treatment. However, when comparing only patients with a baseline HAM-D17 of 18 or more, no differences were seen for HAM-D17 (Table 6) while HAM-D6 still was scored highest in the US trials compared to the EU trials.
Table 4

All patients included in the six trials (US trials = 1,043, EU trials = 576)

HAM-D17*

HAM-D6*

US

EU

US

EU

19.3 (4.7)

20.5 (3.7)

10.8 (2.7)

10.5 (2.2)

*P≤0.05 for geographic differences based on analysis of covariance

Table 5

The subset of patients who maintained the longest duration of therapy with a median duration ranging from 56 to 63 days across the fixed-dose groups

HAM-D17*

HAM-D6

US

EU

US

EU

19.4 (4.7)

20.4 (3.6)

10.8 (2.6)

10.5 (2.2)

*P≤0.05 for geographic differences based on analysis of covariance

US trials N=637 (61.1%) and EU trials N=427 (74.1%)

Table 6

The same subset of patients as in Table 5, but including only those with a baseline HAM-D17 of 18 or more

HAM-D17

HAM-D6*

US

EU

US

EU

21.8

21.6

12.0

11.1

(3.2)

(2.7)

(1.8)

(1.9)

*P≤0.05 for geographic differences based on analysis of covariance

US trials N=432 (41.4%) and EU trials N=349 (60.6%)

The subset of patients who maintained the longest duration of treatment had median treatment durations of 56 days for the duloxetine 40 mg daily group (N=103), duloxetine 80 mg daily group (N=231), and the duloxetine 120 mg daily group (N=144); 59 days for the placebo group (N=417) and 63 days for the duloxetine 60 mg daily group (N=169). Table 7 shows the effect size results by dose for this population. The results are similar to those seen with all patients. Effect sizes >0.40 for each dose group based on the HAM-D6, and the HAM-D6 was more sensitive than the HAM-D17 at detecting differences with the exception of the duloxetine 40 mg daily group.
Table 7

Endpoint effect sizes based on change from baseline to endpoint for the 70% of the study population maintaining the longest duration of treatment

Endpoint HAM-D effect size (with 95% confidence intervals) for duloxetine via LOCF

40 mg daily (N=197)a

60 mg daily (N=350)

80 mg daily (N=467)

120 mg daily (N=286)

HAM-D6

HAM-D17

HAM-D6

HAM-D17

HAM-D6

HAM-D17

HAM-D6

HAM-D17

0.42 (0.13, 0.70)

0.43 (0.14, 0.71)

0.51 (0.30, 0.72)

0.48 (0.27, 0.70)

0.48 (0.30, 0.66)

0.39 (0.21, 0.57)

0.65 (0.41, 0.88)

0.62 (0.39, 0.86)

aThe ratio of the sample size between duloxetine and placebo is ∼1; ranging from 0.93 to 1.10

Table 8 shows the results for the subset of patients described in Table 7, but includes only patients with a baseline HAM-D17 of 18 or more. The duloxetine 40 mg daily group has been deleted because only 54 patients scored 18 or more on HAM-D17 (all US patients). The pattern of the dose–range from 60–120 mg in Table 8 is similar to the one found in Table 7, but the effect sizes are generally lower apart from the one for HAM-D17 for the dose of 80 mg daily, which in Table 8 is above 0.40.
Table 8

Patients with HAM-D17 ≥18 at baseline

Endpoint HAM-D effect size (with 95% confidence intervals) for duloxetine via LOCF

60 mg daily (N=281)

80 mg daily (N=325)

120 mg daily (N=232)

HAM-D6

HAM-D17

HAM-D6

HAM-D17

HAM-D6

HAM-D17

0.43 (0.20, 0.67)

0.41 (0.18, 0.65)

0.49 (0.27, 0.72)

0.44 (0.23, 0.66)

0.56 (0.30, 0.82)

0.58 (0.32, 0.84)

Discussion

This study has confirmed previous dose–response trials with SSRIs (Bech et al. 2002) showing the HAM-D6 is more sensitive to measuring antidepressive effect than the HAM-D17. A clearer dose–response pattern emerged via the HAM-D6 compared with the HAM-D17 from fixed-dose studies of duloxetine ranging from 40 mg daily to 120 mg daily. When using the HAM-D6, the results of this study indicate that duloxetine 60 mg daily is the optimal effective dose in patients with major depression. A clinically significant response (effect size ≥0.40) was seen 2 weeks after initiation of therapy. For higher doses of duloxetine (80 or 120 mg daily), the clinically significant responses were obtained only after 6 weeks of therapy and even at endpoint the effect sizes in these higher doses were not different from the effect size obtained with duloxetine 60 mg daily.

However, when comparing the US trials with the EU trials, we found that approximately 88% of the EU patients had completed the planned acute treatment phase while only around 63% of the US patients were completers. As geography was not considered a randomization factor, we analyzed the subset of patients with the longest duration of treatment. Moreover, when comparing US trials with EU trials, we confirmed the results by Niklson and Reimitz (2001) showing that patients in US trials have lower HAM-D17 baseline scores than patients in EU trials, while in US trials they had higher HAM-D6 baseline scores than in EU trials. These differences were numerically very small, but did reach statistical significance (P≤0.05) because of the large number of observations.

When looking at the subset of patients with the longest duration of treatment, we found that 40 mg duloxetine daily obtained an effect size of just above 0.40. When evaluating within this subset of patients those with a HAM-D17 baseline score of 18 or more, we found (Table 8) that the effect sizes for 80 mg duloxetine daily were higher than those for 60 mg duloxetine daily, both on HAM-D6 and on HAM-D17. In the subset of patients with a HAM-D17 baseline score of 18 or more, those receiving 40 mg duloxetine daily were too small in numbers for an adequate comparison with those receiving 60, 80, and 120 mg duloxetine daily. An equal number of patients in each treatment arm have been found to be an essential condition for pooling data from different trials into a metaanalysis (Pogue and Yusuf 1998). In general, the effect sizes for patients with a HAM-D17 baseline score of 18 or more were not higher than for the whole group of patients. However, further studies are therefore needed to evaluate the optimal dose of duloxetine in severely depressed patients.

The somewhat inverse relationship between dose and rates of discontinuation due to AEs suggests factors other than dose influencing the severity of AEs. Geographic and/or cultural factors may have influenced the reporting of AEs or the severity or tolerability of these AEs such that the expected response was not seen (Burroughs et al. 2002a,b). Regardless, effects size analyses of the subset of patients with the longest duration of treatment were confirmatory with respect to the increased sensitivity of the HAM-D6 compared with the HAM-D17. In addition, these assessments suggest that patients who tolerated longer acute treatment with duloxetine may achieve better benefit at the higher dose of 120 mg daily which demonstrated a 26.3% increase in effect size compared with the 60 mg daily group based on the HAM-D6.

The use of effect size as the most simple outcome statistic in trials of antidepressants was demonstrated in the fluoxetine metaanalysis (Bech et al. 2000). The use of a minimum effect size of 0.40 as an indicator of a clinically significant response was confirmed in a metaanalysis of mirtazapine, in which the HAM-D6 was used to test the pure antidepressive effect of an unselectively acting drug with both serotonin and noradrenalin properties, but also with other properties such as an antihistaminergic effect (Bech 2001).

The validity of a depression-rating scale depends as much on the sample of patients being tested as on the scale itself (Borsboom 2005). The patients included in this analysis fulfilled the DSM-IV criteria for a major depressive episode and were treated in the outpatient setting. The minimum required baseline HAM-D17 score was 15, clearly lower than the inclusion level of 20 in the citalopram dose–response study mentioned previously (Bech et al. 2002). The use of the HAM-D6 to monitor antidepressant treatment in outpatients or in primary care patients with major depression has been seen both in US (O’Sullivan et al. 1997) and European (Bobes et al. 2003; Licht et al. 2005) trials. Paykel (1990) showed that in the primary care setting, the advantage of amitryptiline over placebo was significant in the HAM-D17 pretreatment range between 16 and 24 which closely corresponds to the range in the present analysis. In patients with severe depression, a higher dose than duloxetine 60 mg daily might be considered; however, in patients with mild to moderate DSM-IV major depression, duloxetine 60 mg daily was identified by the HAM-D6 to be the optimal dose.

These results are based on a relatively homogenous population of over 1,600 patients with DMS-IV-defined MDD that is demographically representative of the general adult MDD population. The use of a metaanalytic technique for pooling effect sizes across studies further strengthens the validity of these findings. However, several limitations should be considered. HAM-D6 assessments were conducted post hoc; protocol-defined primary assessments of efficacy in each study were based on comparisons of mean change from baseline in HAM-D17 total score and not placebo-referenced effect size. The use of effect size for dose response, while valid, does not replace the gold standard of a well-controlled dose–response study. However, even when considering these limitations, our results do clearly suggest that doses of 60–120 mg duloxetine are clinically effective, while a dose of 40 mg daily may be suboptimal.

Acknowledgements

The authors wish to thank Ling Ling Xie, MS; Barry M. Brolley, MS; Quan Zhao, MS; and Ying Yuan Chen, MS, for their assistance with statistical analyses and reporting. The clinical studies referred to in this work were conducted under the Declaration of Helsinki and confirmed to all applicable local and region regulations. All protocols and informed consent documents were approved by local or regional review boards before study initiation, and all patients were required to sign an informed consent before participation in any study-related activities.

Copyright information

© Springer-Verlag 2006