Healthy Skepticism International News

February 2006

Duloxetine a serotonin and norepinephrine reuptake inhibitor: Superior to existing SSRIs?

Duloxetine a serotonin and norepinephrine reuptake inhibitor: Superior to existing SSRIs ?

Florence Delamarre Key words: Duloxetine, Depression, Advertising Abstract: Background: Duloxetine is a new antidepressant for the treatment of Major Depressive Disorder (MDD); the manufacturer claims that duloxetine is more effective than SSRIs with a similar safety profile. The purpose of this study is to assess whether these claims are well founded or not. Methods: A systematic literature search of Cochrane, EMBASE and MEDLINE was performed to collect results of duloxetine in clinical trials in Major depressive disorder, either when presented alone or in reviews or pooled analyses a the time of launch in the US (August 2004). Abstracts and posters of trials presented in major scientific meetings or symposia were also included in this research. Efficacy and tolerability data were reviewed and analysed against the claims of the manufacturer. Results: Efficacy of duloxetine was demonstrated against placebo at the recommended 60 mg dose but it was not tested against SSRIs. Comparison of higher doses did not show any significant advantage (less than 1 point on the HAMD scale) while administration of these higher doses resulted in high discontinuation rate due to Adverse Events, weight gain, and cardiovascular problems (tachycardia and hypertension). Collected information from published clinical trials does not provide evidence that Duloxetine is more efficient than other SSRIs or that is has a better safety profile than SNRIs. Conclusion: The manufacturer’s claims on superiority for Duloxetine over SSRIs are unwarranted. Introduction: Cymbalta (duloxetine) is a serotonin and norepinephrine reuptake inhibitor (SNRI) registered for the treatment of major depressive disorder (MDD). The class of SNRIs now comprise three medications: venlafaxine, milnacipran and duloxetine. These 3 drugs block the reuptake of serotonin (5HT) and noradrenaline (NE) reuptake in different degrees, duloxetine being presented as the most balanced reuptake inhibitor between the 2 pathways (Schatzberg 2003). Drugs acting on both pathways are claimed to convey greater efficacy on depression than those acting on a single pathway, impacting emotional and somatic symptoms and leading to remission with no residual symptoms (Detke JPR 2002, Detke JCP 2002). This is supposed to confer a broader and stronger efficacy outcome on SNRIs than on selective serotonin reuptake inhibitors (SSRI), alhtough no pharmacological evidence was found to support this theory. In the same time, activating 2 pathways may also result into more side effects, combining serotonergic and noradrenergic adverse events for SNRIs. However, as a balanced reuptake inhibitor, duloxetine is presented as a drug with good tolerability profile, devoid of the noradrenergic cardiovascular toxicity that is observed with venlafaxine. Therefore, duloxetine is presented as a drug having a better efficacy profile than SSRIsassociated with a SSRI like safety profile. This study aims at to assess whether the claimed efficacy and tolerability profile of duloxetine in the treatment of MDD is well founded or not. Methods: A literature search was performed in August 2004 using a systematic investigation of EMBASE, Cochrane and MEDLINE with the keywords ‘duloxetine, clinical trial and depression’. Relevant references from retrieved articles and reviews in the literature were also searched for further potential articles. Abstracts and posters of trials presented in scientific meetings or symposia were also included in this research. All publications presenting data from randomised, doubleblind (or open label for safety issues) and placebo controlled trials as well as review or pooled analyses of such data were included in the review. Study results were analysed for efficacy and tolerability and compared with claims made by Eli –Lilly. Results : As at the time of licensing (Aug 2004), a total of 8 placebocontrolled, double blind shortterm (8 to 9 weeks) studies (named studies 1 to 8), one doubleblind longterm (26 weeks) relapse prevention study in MDD study (named study 9) and 1 longterm openlabel safety study (named study 10) could be identified from the literature search, all performed by the manufacturer (see table 1). Only 5 of these studies were published in peerreviewed journals (studies 1 Goldstein 2002, 5 Goldstein 2004, 7 Detke JCP 2002 and 8 Detke JPR 2002 as well as study 10 Raskin 2003) while the results of other studies were found in a metaanalysis of all duloxetine data in MDD presented in a poster (Falissard 2003). These 8 shortterm doubleblind studies consisted into 2 phase II trials (with fluoxetine as an active control) and 6 phase III trials (with paroxetine as a control for 4 trials and no active control for 2 trials). Only 2 of these studies were performed at the 60 mg o.d. recommended dose, the other acting as “dosefinding” studies. The longterm doubleblind relapse prevention study (study 9) was also performed at 60 mg o.d. while the openlabel study (study 10) tested the longterm safety of other doses (40 and 60 mg bid doses). An additional openlabel study (named study 11) was also analysed, although not placebo controlled, as it is interesting to show that rapid dose escalation from 60 to 120 mg/d in one daily dose was analysed. For the overall development plan it appears that there is no rational strategy to evidence an effective dose: - A formal escalation dose study is missing to find out the optimal dose - Most studies were performed testing twice daily doses (total daily doses of 40, 80 and 120 mg/day). Few studies tested a once daily dose (60 mg/day), and there was no link with previously tested doses; - These latter studies test the therapeutic dose recommended but do not use active control and do not evaluate the efficacy based on MADRS scale and anxiety based on HAMA scale as the other studies do. Shortterm efficacy General efficacy. Efficacy compared to placebo was recognised by US Food and Drug Administration (FDA) at the dose of 60mg/day. At that dose, duloxetine provided a greater improvement than placebo of 4.9 points in the change of HAMD (in study 7, p < 0.001 from placebo, Detke JCP 2002) and 2.2 points (in study 8, p<0.05, Detke JPR 2002). The corresponding response rates (45% versus 23 % p<0.001 in study 7 and 50% versus 35% p=0.017 in study 8) were significantly greater than placebo, but not systematically for the remission rates (respectively 31 versus 15% for placebo p=0.03 and 32 % versus 24% for placebo, NS). An overview of results issued from all 8 phase II and III studies can be visualised in the exhaustive metaanalysis presented by Falissard: approximately half of the studies evidenced a significant difference between duloxetine and placebo in terms of change in HAMD score (see figure 1), response or remission rate, leading to an overall significant improvement of 2.2 points in the HAMD17 change from baseline compared to placebo. The difference evidenced between duloxetine and placebo highly varied between the studies, the highest differentiation being supported by the studies where no active control was used. Whether efficacy is dose dependent is difficult to establish from the data, and it is not clear whether higher doses (80 to 120 mg/day) would lead to stronger efficacy. Efficacy on anxiety symptoms. The efficacy of duloxetine on anxiety symptoms associated with depression was assessed on - the HAMA (in 4 studies but not in the recommended dose studies), - the HAMD item 10 (anxiety/psychic item) - and anxiety/somatisation subfactor of HAMD Results can be found in individual study publications as well as in a review of 4 studies (Dunner 2003), i.e. one fluoxetine study (n°1), one paroxetine study (n°5) and the 2 recommended dose studies, (n° 7 and 8) that were selected as they were ‘positive’. The evaluation was performed in depressed patients who had no primary anxiety disorder in the year preceding the study. At the recommended dose, duloxetine lead to a significant improvement of the HAMD item 10 and of the anxiety/somatisation subscale in only one of the 2 recommended dose studies (n°7). Results observed on other doses are inconsistent; in one paroxetine study (study n° 5) duloxetine 80 mg significantly differentiated from placebo on all measures (with a 2.2 point difference from placebo on the HAMA score, Goldstein 2004), and in a fluoxetine study (n°1) it significantly differentiated from placebo on the HAMD10 and anxiety/somatisation subscale only (Goldstein 2002. In other studies, duloxetine did not lead to greater improvement compared to placebo (Dunner 2003). Efficacy on physical symptoms. Physical symptoms were assessed through - Item 13 of the HAMD17 scale, “somatic symptoms (general)”, - the 28item Somatic Symptom Inventory (SSI) total score or 7 painrelated Items (items 2, 3, 9, 14, 19, 27, and 28) - Visual Analog Scales (VAS) for overall pain, headache, back pain, shoulder pain, interference with daily activities and time in pail while awake. Efficacy on pain symptoms were not assessed in Phase II trials. Results could be found in few individually published articles (study 3, 7 and 8) and in a pooled analysis of the recommended dose studies (studies 7 and 8, Goldstein Psychosomatics 2004). At the dose of 40 mg/day, duloxetine does not provide any significant benefit compared with placebo on any of the scales used. At higher doses, results of HAMD item 13 and the item SSI were significantly different from placebo in only one study testing 60 mg/d (study n°7). When pooling together results of the 2 recommended dose studies (n°7 and 8), results are no longer significant. Concerning VAS scales, significant results are seen for overall pain at some time points only in both studies (n°7 and 8), on back pain at endpoint in one study (n°7), and on headache and shoulder pain at endpoint when both studies are pooled together. The differences in VAS scores between duloxetine and placebo treatment groups were of approximately the same magnitude at early and late visits, with somewhat larger and sometimes statistically significant differences noted at intermediate visits. Additional statistical analyses were used (median, percentiles, AUC analysis) to provide significant differences with placebo (Goldstein Psychosomatics 2004). However, whatever the statistical analysis used to evidence differences, these advantages do not translate into any perception of benefit for the patient since results for VAS on interference for daily activities and time in pain while awake are never significantly different from placebo. Comparison with efficacy of SSRIs. Duloxetine results were compared a posteriori to those of paroxetine and fluoxetine in the 6 trials where SSRIs were used as active control. Results of comparison can be found in posters pooling the data from all studies versus paroxetine and fluoxetine (Thase 2003, Swindle 2004). First, it should be reminded that: 1) These comparisons are unbalanced in terms of doses: duloxetine tested large ranges of doses (from 40 to 120 mg/day) while paroxetine / fluoxetine were given as fixed 20 mg daily dose. 2) The 60 mg/day recommended dose for duloxetine was not tested against SSRIs and only extrapolation of the study results can be made to have balanced comparisons of the compounds. When focusing on available individual study results, it can be noted that the dose of 40 mg/day did not differentiate from paroxetine (study n° 5, Goldstein et al 2004). It is interesting to notice that this dose was excluded from the pooled analysis as it was “subsequently determined to be suboptimal” (Swindle 2004, Thase 2004). The superior efficacy of duloxetine 80 mg/day that is demonstrated in the published paroxetine trial (Goldstein JCP2004) is based on HAMD17 and remission rate estimated through MMRM. When comparing LOCF results, differences are not significant anymore and their clinical relevance can be questioned: duloxetine 80 mg/day leads to a 1.6 point higher change in HAMD17 and remission rates of 30% for placebo, 37% for paroxetine and 35% and 50% for duloxetine 40 and 80 mg respectively. The unusual aspect of the paroxetine efficacy timecurve with poorer effects at week 8 than at week 6 and week 4, may contribute to this end of study difference. When results of high dose arms of all 6 studies are pooled (i.e. comparison of duloxetine 80 and 120 mg/day to fluoxetine or paroxetine at 20 mg/day), duloxetine presents significantly superior efficacy in terms of HAMD17 mean change (estimated using MMRM) of less than 1 point (Swindle 2004), and a non significant difference between remission rates (38% versus 43% respectively (Thase 2004). 5 The claimed superior efficacy on anxiety symptoms is based on a significantly greater reduction of the MMRM HAMD anxiety/somatisation subscale (with p=0.045), but no significant difference are evidenced for HAMA and HAMD item 10 with the MMRM method (no LOCF results were given). Finally, superiority on the efficacy on physical symptoms of depression is ascertained on shoulder pain only; no difference is evidenced between duloxetine and paroxetine on all other VAS scale for pain. Safety Profile: Information comes from individual studies as well as a pooled analysis of 6 of the 8 trials performed on MDD (Nemeroff CB, 2002) and a pooled analysis comparing duloxetine to SSRIs presented in a poser (Tran 2003). General tolerability. The rate of discontinuation due to adverse events is high for duloxetine, compared both to placebo (15% versus 5% in the pooled analysis of 6 studies from Nemeroff) (n= 1755, p<0.001). In studies testing the recommended 60 mg dose (studies n° 7 and 8), discontinuation rates due to adverse events reach 12.5 and 13.8 % for duloxetine compared to respectively 4.3% and 2.5% for placebo. Higher doses seem difficult to tolerate: in study 5, among patients who began duloxetine 80 mg/d: 22% reduced their dose up to 2 weeks because of adverse events (Goldstein et al 2004). In the phase II fluoxetine study (study n° 1), patients were forcetitrated from 40 mg/d to 120 mg/d in 3 weeks, and were to be administered 120 mg/d unless for safety reasons (Goldstein et al 2002). Up to 25% patients did not receive the 120 mg/dose (but a lower dose). Moreover, 6 duloxetine patients (versus 2 for placebo) discontinued in the first week of treatment (before their first evaluation). Similarly, in the phase II open label study, testing rapid dose escalation from 60 to 120 mg/d over 3 weeks (study n°11), 20 % of the 128 included patients were not able to reach the final 120mg once daily dose (Wohlreich 2004). The most frequently reported adverse events were nausea, dry mouth, fatigue and insomnia. The adverse events responsible for significantly more discontinuation for duloxetine when compared to placebo were nausea (2.4% vs. 0.3%, p<0.01) and dizziness (1.1% vs. 0.1% p<0.05 (Nemeroff 2002). Nausea was reported more frequently at the recommended 60 mg/day dose are up to 30 and 47% of patients (versus 12 and 9% for placebo respectively in studies 7 and 8). In the only published study where both duloxetine and paroxetine were administered, rates of discontinuation due to adverse events were higher under duloxetine (calculated from the CONSORT diagram to 11.6 and 15% of patients for 40 and 80 mg/d respectively) compared to paroxetine (9%) and placebo (9%) (Goldstein 2004). In the overall metaanalysis of all paroxetine studies (Tran 2003) 4% of the placebotreated patients (n=371), 8% of duloxetine (n=736) and 6% of paroxetine (n=359 patients) discontinued due to the adverse events with no statistically significant difference between duloxetine and paroxetine. However, it should be noted that this pooled analysis includes patients receiving duloxetine 40mg/d while the efficacy metaanalysis did not, since 40mg/d was considered a subtherapeutic dose (Swindle 2004). There were no significant difference in treatmentemergent adverse events between duloxetine and paroxetine, except for insomnia (reported by a significantly greater proportion of duloxetine 80mg/d patients than in the paroxetine patients (p=0.03) in study 5 (Goldstein 2004), and for an increased proportion of decreased appetite under duloxetine (p=0.017) in the pooled analysis (Tran 2003)). Withdrawal symptoms. In a paroxetine study (n° 5, Goldstein 2004), discontinuation emergent treatment events for dizziness (20.8%), nausea (11.3%) and paresthesia (11.3%) were reported significantly more frequently by patients receiving duloxetine 80 mg/day than those receiving placebo, and than those receiving paroxetine for paresthesia (0%). In the longterm study (n°10) were duloxetine was stopped abruptly, withdrawal events were dizziness (8.3%), anxiety (4.3%), nausea (4.2%), headache (3.1%), insomnia (2.9%) and irritability (2.6%) Raskin 2003. In a phase I study performed in 8 healthy volunteers (a 3 weeks study with increasing doses from 20 to 40 mg/d) abrupt discontinuation of the drug was also associated with increases in heart rate of 4.5 bpm (1.9; 11.8) in the recumbent position and 4.5 bpm (1.3; 9.7) in the standing position (Sharma 2000). Cardiovascular Safety. In the long term treatment (study n° 10), administration of duloxetine 80 to 120 mg/day was associated with an heart rate increase (significant increase of 2 bpm in average, Raskin, 2003). More worrying were increases in blood pressure which were limited when presented as average value (1.5 to 2 mmHg increase in SPB and DPB in average), but seemed to affect more importantly some patients: up to 4% of patients under longterm treatment developed a sustained hypertension which remained non reversible for almost half of them: 2 % of the treated patients did not return to baseline blood pressure after treatment discontinuation (Raskin 2003). In the pooled analysis comparing paroxetine and duloxetine (Tran 2003), changes from baseline to endpoint of vital signs reach statistical significance for supine hear rate (p<0.01). It was stated that these changes were not clinically meaningful. However, it can be noted that a higher proportion of patient under duloxetine (around 2%) develop sustained hypertension within these shortterm trials, compared to paroxetine (around 0.5%). The difference is nto statistically significant and a similar proportion of patients also developed sustained hypertension under placebo (Tran 2003). Discussion Claims for broader and stronger efficacy According to the dual action concept accompanying duloxetine development, a compound acting on two different pathways is superior to compounds acting on a single pathway in terms of suiting a broader population (Delgado 1993), greater efficacy, and acting on both emotional and physical symptoms of depression. If venlafaxine is now recognised to be superior to SSRIs (Thase 2001), is that attributable to the dual pharmacological action of the compound? Is that also the case with duloxetine? Efficacy compared to placebo was recognised by US FDA at the dose of 40 and 60mg/day although efficacy of the 40 mg/day can be questioned. EliLilly recommends increasing to 60mg/day as soon as possible (ref). At therapeutic doses, duloxetine did not demonstrate clear and consistent efficacy on either anxiety or physical symptoms associated with depression. Efficacy on anxiety was notassessed by the standard HAMA scale at the recommended 60 mg/day dose. Results were negative at 40 mg/day and inconclusive at higher doses. Efficacy on anxiety was then claimed by EliLilly based on subanalyses of the HAMD scale which are known to provide less convincing evidence of efficacy on anxiety. Evaluation of efficacy on physical symptoms, based on a specific pain scale (SSI) and nonspecific VAS scales, resulted mostly in non significantly different results compared to placebo. The use of additional statistical approaches to evidence such differences, (présenté avant) brings confusion to the overall results, which are not convincing. Positive results observed for overall pain, back pain and shoulder pain at some time points do not translate into a superior relief from interference with daily activity or pain while awake. The superior efficacy to SSRIs claimed by EliLilly, is extrapolated from the comparison of high doses of duloxetine (80 and 120 mg/day) to low doses of paroxetine and fluoxetine. At the 40 mg/day dose, duloxetine does not result in a different efficacy when compared to SSRIs. In the study testing 80 mg/day of duloxetine versus paroxetine, significance on the primary endpoint disappeared when the LOCF measure (which consider drop out rates) was used. When studies were pooled, the comparison results in significant difference for both HAMD score and remission over SSRIs. It is important to evaluate the clinical relevance of the difference; with less than 1 point difference in the HAMD total score and 5% in remission rates, the superiority remains questionable. Inference from negative results at 40 mg/day and some clinically dubious positive results at 80 mg/day, leaves us unsure that Eli Lilly’s recommended dose of 60 mg/day of duloxetine would present any superiority in efficacy to SSRIs. The differentiation observed at doses of 80 and 120 mg /day, suggesting a better efficacy of the compound at these higher doses did not convince the FDA, which noted “there is no evidence that doses greater than 60 mg/day confer any additional benefit”. A SSRIlike tolerability profile? EliLilly positions duloxetine as an antidepressant with higher efficacy profile associated with a “SSRIlike tolerability profile”, different therefore from its main competitor venlafaxine. However, the available data on the tolerability do not support this claim. Duloxetine presents a high discontinuation rate for adverse events (3 times that of placebo) significantly higher than observed with SSRIs in the direct comparison studies (p<0.05). Withdrawal symptoms affect patients on duloxetine more than paroxetine does, suggesting that withdrawing from duloxetine treatment might represent a non negligible issue. Finally, as expected from a drug acting on the noradrenergic pathway, the cardiovascular profile of duloxetine might not be so inoffensive as presented, with increased heart rate and blood pressure (sometimes non reversible) observed at 80 or 120 mg/dose. The tolerability profile presents duloxetine more as a venlafaxine like compound than a SSRIlike compound. Conclusion The available data on duloxetine do not support the claim of a broad efficacy on physical and emotional symptoms, or of higher efficacy when compared to SSRIs associated with similar tolerability profile. The benefitrisk ratio of duloxetine appears to be weak and not superior to existing antidepressants. At low doses it is ineffective; at higher doses it produces unacceptable side effects, and it has a worrying withdrawal profile. For those patients who may need higher than recommended doses, we strongly advise against increased doses of duloxetine, as potential cardiovascular events may occur, notable after longterm treatment.

STUDY	PUBLICATION	DESIGN	REGIMEN
Short term double blind studies
1	Full publication Goldstein DJ et al, J Clin Psychiatry 2002	Phase II Randomized, controlled doubleblind (n=173)	Duloxetine 60mg bid vs Fluoxetine 20mg o/d vs placebo x 8 weeks
2	Not published (efficacy results presented in a metaanalysis poster)	Phase II Randomized, controlled doubleblind (n=194)	Duloxetine 60mg bid vs Fluoxetine 20mg o/d vs placebo x 8 weeks
3	Not published (efficacy results presented in posters)	Phase III Randomized, doubleblind	Duloxetine 40 mg bid vs Duloxetine 60 mg bid vs paroxetine 20 mg o/d vs placebo x 8 weeks
4	Not published; (efficacy results presented in posters)	Phase III Randomized, doubleblind	Duloxetine 40 mg bid vs Duloxetine 60 mg bid vs paroxetine 20 mg o/d vs placebo x 8 weeks
5	Goldstein et al, J Clin Psychopharmacology 2004	Phase III Randomized, doubleblind (n=353)	Duloxetine 20 mg bid vs Duloxetine 40 mg bid vs paroxetine 20 mg o/d vs placebo x 8 weeks
6	Not published (efficacy results presented in posters)	Phase III Randomized, doubleblind	Duloxetine 20 mg bid vs Duloxetine 40 mg bid vs paroxetine 20 mg o/d vs placebo x 8 weeks
7	Detke et al, J Clin Psychiatry 2002	Phase III Randomized, doubleblind (n =245)	Duloxetine 60 mg o/d vs Placebo for 9 weeks
8	Detke et al, J Psychiatry Res 2	Phase III Randomized, doubleblind (n =267)	Duloxetine 60 mg o/d vs placebo for 9 weeks
Other studies
9	Not published (mentioned in posters)	Phase III Randomised, doubleblind, relapse prevention	Duloxetine 60 mg o/d vs placebo for 26 weeks (after 12 weeks in open label)
10	Raskin J et al, J Clin Psychiatry 2003	Phase III Randomised, openlabel, placebo controlled	Duloxetine 80 or 120 mg per day for 52 weeks
11	Not published (presented in posters)	Phase II Randomised, openlabel	Duloxetine escalating doses from 60 to 120 mg o/d

 

View/Hide References

 

HS Int News index

Page views since 15 March 2010: 9138

 

Comments

Our members can see and make comments on this page.