SummaryThe drug and the review Agomelatine is a new antidepressant with selective agonist actions at mela-tonin receptors and

d selective antagonist action at serotonin 5HT-2C recep-tors. It do es not affect the uptake of serotonin, noradrenaline or dopa-mine. Agomelatine was launched in the UK in June 2009. This review evaluates the evidence supporting agomelatine use in adultswith majo r depressive episodes and seeks to define agomelatines potentialplace in therapy. Background NICE Guidance (2004 and draft update 2009) recommends an SSRI antide-pressant (e .g. generic fluoxetine or citalopram) first-line for patients withmoderate to se vere depression. Following first-line treatment failure the evidence for sequencing antide-pressa nts is inconclusive, choice should therefore be based on the needs of the patien t, the profile of the drug and financial considerations. New NICE Guidelines on the treatment of depression in adults are antici-pated in September 2009. Literature searched We searched: Medline (agomelatine.af [Limit to: Humans and English Lan-guage]); Embase (agomelatine.af [Limit to: Human and English Language]and [DEPRESSION/dt [Drug Therapy] or *MAJOR DEPRESSION/dt [DrugTherapy]]); and IDIS ("AGOMELATINE 2 8160472"). This was supplemented with information from the EMEA website (PublicAssessment R eport), NICE guidance, and contact with the manufacturerServier Laboratories. The study programme was extensive: 1 short-term placebo-controlled dose ranging study (8 weeks) 2 short-term placebo-controlled studies (6 weeks with optional exten-sion for a further 46 weeks) 3 short term placebo-controlled studies with paroxetine or fluoxetineactive contr ols (6 weeks and optional extension to 6 months) 2 long-term relapse prevention studies. 1 study vs. venlafaxine (6 weeks with optional extension to 6 months),primary en dpoint sleep, secondary endpoint efficacy 1 study vs. venlafaxine (12 weeks with optional extension to 6months), primary e ndpoint sexual function, secondary endpoint effi-cacy 1 study vs. sertraline (6 weeks with extension to 6 months), primary endpoint ac tigraphy,secondary endpoint efficacy 1 open study in patients with major depressive disorder (6 weeks) looking at the effect onsleep EEG

1 study vs. fluoxetine in patients with severe depression (8 weeks with optional extension to6 months) 1 study on sexual disturbance vs. paroxetine. Efficacy studies Antidepressant efficacy is generally determined from 68 week RCTs comparing respo nseagainst placebo or active comparators, or 610 month RCTs comparing relapse rat e against pla-cebo; depression symptoms are evaluated using rating scales. All analyses were carried out in the intention-to-treat populations: i.e. analys is was performedaccording to the assigned treatment group regardless of protocol deviations and participantcompliance or withdrawal. Short-term placebo-controlled A dose-ranging study (n=711) identified agomelatine 25mg daily as the target dos e when com-pared with agomelatine 1mg and 5mg daily In two 6 week placebo controlled RCTs patients with a HAMD score 22 [moderate to severedepression] were randomised to either agomelatine (25mg in itial for 2 weeks increased to50mg in non-responders) or placebo; primary endpoi nt was mean final HAMD score. In the first study, 212 patients received agomelatine (n=106) or placebo (n=105) . Forthe ITT population the between group difference for the mean final HAMD sco res was2.30 (S.E. 1.02), p=0.026. In the second study, 238 patients received agomelatine (n=118) or placebo (n=120 ).For the ITT population the between group difference for the mean final HAMD sc ores was3.44 (S.E. 0.92), p<0.001 3 unpublished placebo-controlled studies in which paroxetine or fluoxetine were used as activecontrols to validate the study design are reported in the EMEA pub lic assessment report (EPAR).The efficacy of agomelatine was not directly compar ed to that of the active controls.

All 3 studies enrolled patients with moderate to severe depression [HAMD 22] and fol-lowed similar methodologies: initial run-in followed by randomisatio n to agomelatine,placebo, or active control for 6 weeks with subsequent 18 week extensions. The option to increase the agomelatine dose from 25mg to 50mg was not included i nthese studies. The primary endpoint of difference in HAMD was compared between active control a ndplacebo, and between agomelatine and placebo. For the first trial, a statistic ally signifi-cant difference ( p =0.008) was seen between active control and placebo at 6 weeks forreduction in H AMD score versus placebo, but not for agomelatine and placebo; Neitherthe second nor third trial showed statistically significant differences between agomelatin eor the active control and placebo for any comparison at either 6 or 24 weeks, m eaningthat no discernable results could be drawn from these studies. A meta-anal ysis of agomelatine trials shows that 23.5% of patients require a 50mg dose, whi ch is a reasonwhy agomelatine may have failed to differentiate from placebo in t he first study. Thesecond and third studies were associated with high placebo re sponse rates. Active comparator studies A 6 week study randomised 313 patients with a HAMD score 22 to agomelatine 2550mg(n=154) or sertraline 50100mg (n=159) [for each drug the l ower initial dose was increasedafter 2 weeks for non-improved patients]. The pri mary endpoint of the study was the efficacyon rest-activity circadian rhythms. F or the secondary endpoint of an improvement in HAMDscore, the difference in scor es after 6 weeks was 1.68, in favour of agomelatine (p=0.031). There were two studies in patients with moderate to severe depression in which a gomelatinewas compared with venlafaxine. In the 12 week study patients were randomised to agomelatine 50mg (n=137) or ven la-faxine 150mg (n=140). The primary endpoint of the study was deterioration in sexual func-tion and demonstrated statistically significant differences in favou r of agomelatine versusvenlafaxine. For the secondary endpoint of the final MADR S score, agomelatine was shownto be at least as effective as venlafaxine (10.1 v s. 9.8). In the 6 week study patients were randomised to either agomelatine (n=165) or ve

nlafaxine(n=167). The primary endpoint was the effects on sleep variables and de monstrated statis-tically significant differences in favour of agomelatine versu s venlafaxine; final HAMD scorewas the secondary endpoint. The final HAMD scores were similar between the groups, indi-cating similar efficacy for agomelatine a nd venlafaxine (9.9 vs. 11.0). A 8 week study randomised 500 outpatients with HAMD score 25 and CGI severity of illness score 4 (severely depressed) to agomelatine 25-50mg (n=247) or fluoxetine 20-40mg (n=2 57) [foragomelatine the initial lower dose was increased at 2 weeks in non-impro ved patients, for fluoxet-ine the initial lower dose was increased at 4 weeks]. The primary endpoint was improvement inHAMD score from baseline, the difference in scores at week 8 was 1.49, 95% CI=[0.20;2.77](p=0.024) in favour of agomelati ne. Long-term relapse-prevention Relapse is classified as the appearance of the symptoms of the index episode soo n after medica-tion is stopped, whilst recurrence is the appearance of symptoms in a new episode. Relapse indi-cates that the treatment duration was too short. A study reported in the EPAR had an 8-week open label phase (n=610) during which response toagomelatine was determined, followed by a 26-week double-blind rando mised phase (n=367;agomelatine 25mg, n=187; placebo, n=180). At the end of the 26 week double-blind phase, a significant effect was not demon strated forthe primary endpoint of relapse (agomelatine 25.9% vs. placebo 23.5%) . The placebo re-lapse rate in this trial was unexpectedly low (approximately ha lf that seen in other studies of similar design). This may be reflective of some methodological aspects e.g. the broad rangeof the severity of depression of pat ients at inclusion. 169 severely depressed patients (n=89 agomelatine, n=80 placebo) (HAMD >25 and C GI-S 5) continued in an 18 week extension; at the end of this extension there was a s tatisticallysignificant difference in favour of agomelatine (21.3% vs. 31.3%) p= 0.046. A second (published) study, that included a high proportion of patients with mod

erate to severedepression reflecting the types of patient generally seen by psyc hiatrists, had an 8-10 week open-label dose determination phase (n=492) followed by a 24 week double-blind phase (n=390) For the primary endpoint of relapse, results at 24 weeks were significantly lowe r (p=0.0001)for agomelatine (21.7%) than for placebo (46.6%); agomelatine reduce d relapse risk by54% (HR 0.458; 95% CI 0.3050.60). 190 patients (n=106 agomelatine, n=84 placebo) continued in double-blind 20 week exten-sion. A statistically significant difference in the time to relapse was d etected: at 10 monthstwice as many patients treated with placebo relapsed (49.9% ) compared to those treatedwith agomelatine (23.9%) (p<0.0001). The three short term studies with active controls all had 18 week extensions. In one study therelapse rates were lower with agomelatine (14.3%) and fluoxetine ( 17.8%) than with placebo(33.3%), p=0.017 and p=0.045 respectively. In the other 2 studies there were no differencesseen between agomelatine or active control an d placebo which means no discernable results couldbe drawn. Safety Agomelatine does not cause weight gain, has a low risk of sexual dysfunction, lo w incidence of gastro-intestinal reactions, absence of discontinuation symptoms, no sedation or daytime drowsi-ness and an overall incidence rate of adverse eve nts that is similar to placebo. Emergent elevations of liver transaminase enzymes more than 3x upper limit of no rmal (3x ULN)occur rarely (approx. 1% of patients). Overall incidences of elevat ions >3xULN in all patients (i.e.regardless of their transaminase values at base line) were 1.1% with agomelatine 25mg and0.72% with placebo (p=not significant). The 0.4% absolute difference with placebo compares to a 0.7% absolute difference seen with duloxetine/placebo. Agomelatine should not be prescribed in patients with cirrhosis or active liver disease. Check liver function tests (LFTs) on initiation; then at around 6, 12 and 24 wee ks; and thereafterwhen clinically indicated (if transaminases are elevated repea t LFTs within 48 hours; if transami-nases are >3x ULN, discontinue agomelatine). Agomelatine should be used with caution in any patient who drinks substantial qu antities of alcoholor is treated with other medicines associated with a risk of hepatic injury.

 Agomelatine should not be prescribed in patients taking potent CYP1A2 inhibitors , such as fluvox-amine and ciprofloxacin, as this may result in increased serum levels of agomelatine. ModerateCYP1A2 inhibitors, such as propranolol, should be used concomitantly with caution. Critical evaluation of the evidence baseUse of depression rating scales HAMD and MADRS measure similar quantifiable elements of depression; CGI measures clinicianrated global improvement and is thus less sensitive Rating scales may not capture functional impairment appropriately despite this b eing an importantdeterminant of depression severity NICE suggest that clinical efficacy is demonstrated by a weighted mean between-g roup differenceof at least three points or a standardised mean difference of at least 0.5 in HAMD rating score EMEA guidance states that acceptable scales for use as the primary endpoint in t rials of new anti-depressants to determine symptomatic improvement include the H amilton Depression rating scaleof Depression (HAMD) or the Montgomery Asberg Dep ression Rating Scale (MADRS). The HAMDscale (17-item scale as recommended) was u sed for the majority of the agomelatine trials; othersused the MADRS. The placebo effect in depression studies Use of a placebo in an antidepressant trial does not necessarily mean that the p atient is untreated.Patients will have regular visits to their doctor, supportiv e help and interest in their welfare. Insome trials they will be able to contact the therapist when they need to. Patients in the placeboarm will receive everyt hing apart from the active drug, and this can constitute a treatment in itself.I t has been estimated that 30% of patients assigned to placebo will respond withi n six weeks. Despite this, SSRIs and serotonin/noradrenaline reuptake inhibitor (SNRI) antide pressants remainsuperior to placebo in treating depression regardless of its sev erity. Study population selection (and enrolment of severely depressed patients) is par ticularly importantfor external validity of antidepressant trials. Particular considerations for agomelatine

In the two short-term placebo-controlled studies, statistically significant impr ovement in symp-toms of depression was seen with agomelatine compared with place bo. When compared with other antidepressants, agomelatine was more efficacious than sertraline andfluoxetine, and as effective as venlafaxine in treating the sympto ms of depression. Agomelatine demonstrated significant efficacy in only one of the two placebo-con trolled relapse-prevention studies. In the first study the relapse rate in the p lacebo arm was unexpectedly lowand may reflect some of the methodological aspect s, such as the broad range of the severity of depression of patients included in the study. A more severely depressed cohort was enrolled inthe second trial, wh ich demonstrated benefit (80% of enrolled patients in the second study vs.46% in the first study). Overall, the data demonstrate a potential role for agomelatine in treating adult s with major de-pressive disorder. The efficacy studies suggest comparable effic acy to currently licensed antide-pressants. Potential benefits over existing therapy As with any new antidepressant, agomelatines place in therapy is not yet clear. A gomelatinemay be useful in the case of failure of another antidepressant, either because of lack of effi-cacy or due to poor tolerability (e.g. unacceptable sid e effects). It has demonstrated superiorefficacy to Sertraline in one study, flu oxetine in another study and demonstrated benefits overvenlafaxine in two other studies. Abrupt withdrawal of agomelatine (after 12 weeks of therapy) has not been shown to be associ-ated with discontinuation symptoms, which can be problematic with o ther antidepressants. Thismeans that no dose tapering is necessary on treatment discontinuation. There is suggestion that sexual dysfunction caused by agomelatine is less than t hat caused byvenlafaxine The pooled analysis of studies using the Arizona Sexual Experience Scale (ASEX)showed that agomelatine was not associated with sexual d ysfunction. In healthy volunteersagomelatine preserved sexual function in compar ison with paroxetine.

Agomelatine had neutral effect on body weight, heart rate and blood pressure in clinical studies.

Sleep studies have shown that agomelatine increases sleep efficiency and slow-wa ve sleep andhas significantly better effects on subjective sleep variables than venlafaxine. Agomelatine has a good tolerability profile compared with other classes of antid epressants. Potential disadvantages over existing therapy There is a lack of long term active comparator controlled studies. As such defin ing a place intherapy in comparison with the other better established alternativ es is not possible. For a newantidepressant agomelatine has robust data for both short and long term treatment effective-ness, which will need to be validated i n real life clinical practice. Health economics

No health economic studies of agomelatine have been published. Budget impact model: per 100,000 population

The budget impact model is based on the use of agomelatine as second-line or lat er treatment.The incidence of this is estimated at 0.40%, affecting 400 patients per 100,000. If agomelatine takes 10% of patients from each second-line treatment, 40 patient s would betreated with it, at a cost of 38.53 per 28 days. Over the first year, taking into account a reduction in prescribing costs of the other second linetreatments and the need for liver function tests, the net incr ease in prescribing costs is esti-mated to be 5,892 (1.45% increase in spend). Th is would rise to an increase of 9,285(2.29% increase in spend) in the second year and 11,607 (2.87% increase in spend) in thethird year, as more patients are pres cribed agomelatine. Background Agomelatine was launched in the UK in June2009 for the treatment of major depres siveepisodes in adults. 1 The recommended dose is25mg, taken at bedtime, which can be in-creased to 50mg a

fter two weeks if there is noimprovement in symptoms. Treatment shouldcontinue f or at least six months to ensure thatthe patient is free of symptoms. No dose ta -pering is required on treatment discontinua-tion. 2 Current NICE guidance (Dec 2004) 3 and thedraft 2009 Guidance 4 both recommend that anSSRI should be used first line in patients withmoderate to severe depression, as they are aseffective as tricyclic antidepressants and les slikely to be discontinued due to side effects.Generic versions, such as fluoxet ine, paroxet-ine, sertraline and citalopram, would be rea-sonable choices. Follo wing first line treatmentfailure (lack of response, withdrawal due toadverse eff ects) the evidence for treatmentsequencing is weak. Choice should thereforebe ba sed on the profile of the drug, needs of the patient and financial consideration s. NICEGuidelines on the treatment of depression inadults are anticipated in Sep tember 2009. Pharmacology Agomelatine is a synthetic analogue of the hor-mone melatonin, and strongly bind s to andstimulates the activity of melatonin MT1 andMT2 receptors, normalising d isturbed circadianrhythms and disrupted sleep-wake cycles. 5 (Melatonin regulates circadian rhythms, includ-ing sleep-wake cycles.) Disturba nces in cir-cadian rhythms have been implemented in thedevelopment of mood disor ders. 5 Agomelatineis also a serotonin-receptor antagonist andbinds to and inhibits the activity of serotonin5HT2C receptors. 5 This antagonism is associ-ated with antidepressant and anti-anxiety ac-tivity, a nd increases slow-wave sleep.Agomelatine does not affect the uptake of se-rotoni n, noradrenaline or dopamine. The inhi-bition of 5HT2C receptors increases norad rena-line and dopamine in the frontal cortex andmay contribute to agomelatines an tidepres-sant activity. 5 Agomelatine has no effect onmonoamine uptake and no affinity for a, badrenergic, histaminergic, cholinergic, dopa-minergic and benzodiazepine receptors. 2 No difference between dosing, effectiveness orsafety of agomelatine between olde r and youngeradults has been reported from trials.

5 There areno data on the use of agomelatine in children. Nospecific data have bee n reported on the safety of it during pregnancy, though animal studies haverepor ted no risks. 5 Agomelatine should be usedcautiously in patients with liver impairment, butits l ack of toxicity and broad safety margin sug-gest lower doses could be used with appropriateclinical monitoring. 5

Special populations Agomelatine is not recommended for use in chil-dren and adolescents below 18 yea rs of age andshould be used with caution in the elderly ( 65years of age) due to lack of clinical efficacy data.There is limited clinical data on the use of agome-latine in patents with moderate or severe renalimpairme nt; it should be used with caution insuch patient. 2 Agomelatine should not be used in patients withhepatic impairment (i.e. cirrhos is or active liverdisease). 2 There is no clinical data on the effect of agome-latine on pregnancy. Animal st udies do not indi-cate direct or indirect harmful effects to the foe-tus. Cautio n should be exercised when prescrib-ing to pregnant women. 2

Safety Agomelatine does not cause weight gain, has alow risk of sexual dysfunction, low incidence of gastro-intestinal reactions, absence of discon-tinuation symptoms and an overall incidence rateof adverse events that is similar to placebo. 6 In clinical trials, over 3900 patients have receivedagomelatine. The incidence of emergent eleva-tions of liver transaminase enzymes more than 3xupper limit of normal (3xULN) was 0.8% in pa-tients treated with agomelatine and with normalba seline levels and 0.3% those treated with pla-cebo. In patients treated with ago melatine 50mg,the incidence was 1.3%. 6 The incidences in allpatients (i.e. regardless of their transaminase val-ues at baseline) were 1.04% (agomelatine 25mg),1.39% (agomelatine 50mg) and 0.72% (plac

ebo)the differences from placebo are not statisticallysignificant. 6 Liver function tests (LFTs) should be performed atthe start of treatment then p eriodically around 6,12 and 24 weeks and thereafter when clinicallyindicated. 2 If serum transaminases rise, the LFTs should be repeated within 48 hours. Therap yshould be discontinued if the increase exceeds3xULN and liver function tests sh ould be per-formed regularly until they return to normal. 2 If any patient develops symptoms which suggesthepatic dysfunction, LFTs should b e carried outand the decision to continue therapy should be aclinical one, based on laboratory evaluations. If jaundice develops, treatment must be stopped.Ago melatine should be used with caution in pa-tients who consume a large quantity o f alcohol orwho are treated with other drugs that may causehepatic damage. 2 Agomelatine is metabolised mainly by cytochromeP450 1A2 (CYP1A2) (90%) and by C YP2C9(10%). Treatment with agomelatine is contra-indicated in patients who are a lso taking potentCYP1A2 inhibitors, such as fluvoxamine and cipro-floxacin as th ey can increase the serum levels of agomelatine. Use of moderate CYP1A2 inhibito rs,such as oestrogens and propranolol with agome-latine should be with caution u ntil more experi-ence has been gained. 2 Agomelatine does notinduce CYP450 isoenzymes and will not modifyexposure to medi cinal products metabolised byCYP450. 2

The placebo response Over the years there has been an increase in thepercentage of patients respondin g in the placeboarms of published studies of major depression. 3 There have been no studies comparing placebowith no treatment in depression, so these patientsrepresent several different kinds of treatment ef-fect, such as p atients who may have had theirdepression scores inflated to ensure study entry,p atients who were accurately rated but show anearly spontaneous remission and tru e respondersto placebo. 3 The difference in improvements indepressive symptoms between the drug and thepla cebo is thought to be greater with increasingdegrees of severity of depression. This is oftendifficult to demonstrate as data is quoted for en-tire groups, rath er than individual patients andthe distribution of the depression rating scoresc an overlap, thereby diluting the effect.

3 Use of a placebo in an antidepressant trial doesnot necessarily mean that the p atient is un-treated. 4 Patients will have regular visits to theirdoctor, supportive help and interest i n their wel-fare. In some trials they will be able to contactthe therapist when they need to. Patients in theplacebo arm will receive everything apart fromthe a ctive drug, and this can constitute a treat-ment in itself. It has been estimate d that 30% of patients assigned to placebo will respond withinsix weeks. 4 The problem for the evaluation of antidepres-sants is that the study population is likely to de-termine the magnitude of the treatment benefit.A poorly selecte d study population will make adrug look bad; a well selected study populationwil l make a drug look better. Since treatment of real patients is not subject to th e perverse in-centives of drug trials, it is often assumed thattrials that recru it a more severe study popula-tion corresponds better to everyday practicethan t rials that do not. The existence of poorlyconducted trials showing minimal benef its can-not be ignored, however, and has been used tocase doubt on the efficacy of antidepressants asa class. Efficacy studies A number of studies have been carried out as-sessing the efficacy and safety of agomelatine.These are described in detail in this review, andtheir design and ma in results are summarised inAppendix 1. All patients enrolled had moderateto sev ere depression. The study programmewas extensive: 1 short-term placebo-controlled dose rangingstudy (8 weeks) 7

2 short-term placebo-controlled studies (6weeks with optional extension for a fu rther46 weeks) 8;9

3 short term placebo-controlled studieswith paroxetine or fluoxetine active contr ols(6 weeks and optional extension to 6months) 6

2 long-term relapse prevention studies 6;10;11

1 study vs. venlafaxine (6 weeks with op-tional extension to 6 months), primary end-point sleep, secondary endpoint efficacy 12

1 study vs. venlafaxine (12 weeks with op-tional extension to 6 months), primary end-point sexual function, secondary endpointefficacy 13

1 study vs. sertraline (6 weeks with exten-sion to 6 months), primary endpoint a ctigra-phy, secondary endpoint efficacy 14;15

1 open study in patients with major depres-sive disorder (6 weeks) looking at th e effecton sleep EEG 16

1 study vs. fluoxetine in patients with severedepression (8 weeks with optional extensionto 6 months) (conference abstract only) 1 study on sexual disturbance vs. paroxeti-ne 18

Guidance from the EMEA 19

for the clinical investi-gation of medicinal products for the treatment of depre ssion states that acceptable scales for useas the primary endpoint to determine sympto-matic improvement include the Hamilton Depres-sion rating scale of Depres sion (HAMD, 17-itemscale) or the Montgomery Asberg Depression Rat-ing Scale. NIC E recommend that where aweighted mean difference is calculated, a be-tween-group (drug/placebo) difference of at leastthree points (two for treatment-resistant depres-sion) is considered clinically significant for theHAMD. 3 The EMEA also state that the disordershould be classified according to an intern ationallyacknowledged classification system, preferablythe DSM-IV: the clinical trials enrolled patientswho did have depressive disorder according to theDSM-IV criteria. Short term studies NICE Guidance recommends that when treating apatient with moderate-severe depres sion, the an-tidepressant dose should be titrated to the recog-nised therapeutic dose and the efficacy of thisdose assessed over 4-6 weeks. 20 If this is effec-tive, treatment should be continued for a further4-6 months, or longer in patients with recurrentdepression, at the full treatment dose. 3;20 If thedose is not effective, it should be increased in linewith the schedule rec ommended in the Summaryof Product Characteristics. If there has been apartial re sponse after a month, a decision toswitch to another antidepressant can be postponed until 6 weeks of therapy. A poorly toler-ated antidepressant should be swi tched to an-other one. 3 The following trials were long enoughto assess the initial efficacy of the treat ment, buttoo short to make any conclusions about whetheragomelatine is effective in maintaining remission.Randomised, double-blind comparisons versusplacebo are needed to permit adequate evaluationof efficacy and for distinguishing disease manifes-tations from adverse reactions. 19 The use of aplacebo is controversial and precautions to mini-mise the impact of the study should be taken:generally a duration of 6 weeks should be suffi-cient and a longer duration should be justified.Three-arm trials including both a plac ebo and ac-tive control are recommended.Nine short term studies have been conduc ted.What these short term studies show is that aftersix weeks of treatment agome latine has a favour-able effect on the symptoms of depression, asseen by reducti ons in HAMD scores and high re-sponder rates. The agomelatine/placebo differ-enc e in HAMD score was not always greater than3 points (ref 8 and the active contro l studies),the difference considered to be clinically signifi-cant by NICE. The agomelatine/sertraline HAMDscore difference was greater than 1.5 points,which is considered clinically significant. Thedose ranging study and the three short te rmstudies with active controls did not have the op-tion to increase the agomelat ine dose to 50mg,which will results in an underestimate of the effi-cacy of agom elatine as ~23.5% of patients withrequire a 50mg dose.Longer term studies will s how whether these effects are sustained over a suitable treatment period of 4-6

months.