ABSTRACT FROM: Zhou X, Ravindran AV, Qin B, et al. Comparative efficacy, acceptability, and tolerability of augmentation agents in treatment-resistant depression: systematic review and network meta-analysis. J Clin Psychiatry 2015;76:e487–98.
Several strategies have been proposed for patients with treatment-resistant depression (TRD), including dose optimisation, switching to another therapeutic class or combination/augmentation regimens,1 however, it is still not known how the different strategies compare with each other. Zhou and colleagues specifically examined augmentation strategies for efficacy and acceptability in a network meta-analysis of randomised controlled trials (RCTs).
PubMed, the Cochrane Library, clinical trial registries and drug company websites were searched for published/unpublished RCTs comparing augmentation strategies (or placebo) for the acute treatment of adults with TRD (defined as major depressive disorder with previous failure and current non-response to at least one oral, first-line, therapeutically dosed antidepressant drug). Original study authors and drug manufacturers were contacted. RCTs were excluded if patients had psychotic symptoms or serious medical comorbidity. Primary outcomes were response and all-cause discontinuations at 6 weeks (or nearest time point between 2 and 14 weeks), and secondary outcomes were remission and discontinuation due to adverse events (AEs). Conventional definitions were applied for response and remission. ORs and 95% CIs for primary and secondary outcomes were first estimated in direct, pairwise meta-analyses (random effects model employed if significant heterogeneity was found). Then, a Bayesian random effects network meta-analysis was carried out with ORs and 95% credible intervals (CrIs). Cumulative probabilities were estimated as the surface under the cumulative ranking curve (SUCRAs). Direct and indirect OR estimates were compared for significant disagreement. Sensitivity analyses examined the effects of treatment dose, study duration, trial blinding, use of data imputation, magnitude of placebo response and other potential confounders. Meta-regression examined the effect of sponsorship.
Forty-eight RCTs (n=6654; mean duration, 6.2 weeks) were identified. Overall study quality was good, but few studies met criteria for low risk of bias, probably due to poor reporting rather than poor study conduct.
Quetiapine, aripiprazole, thyroid hormones (T3/T4) and lithium outperformed placebo for response (ORs ranging between 1.56 and 1.92; 45 studies in the network). T3/T4, risperidone, quetiapine, buspirone, aripiprazole and olanzapine outperformed placebo for remission, with an OR between 1.79 and 2.94 (31 studies). No active treatment was superior to another for either response or remission.
No drug was significantly worse than another active drug or placebo in terms of all-cause discontinuation (44 studies). Quetiapine, olanzapine, aripiprazole and lithium had an increased risk of discontinuation due to AEs relative to placebo (ORs ranging between 2.30 and 5.64; 39 studies).
Statistical heterogeneity was mostly low; publication bias appeared likely only for the quetiapine versus placebo comparison. Direct and indirect OR estimates of primary outcomes did not differ significantly, implying consistency across the network of comparisons.2 According to SUCRAs, no drug was found to be superior to another. The meta-regression analysis found negligible effects for study sponsorship.
Most sensitivity analyses found stronger primary outcome advantages for aripiprazole and quetiapine than for T3/T4 and lithium.
About half or more of the lithium and T3/T4 trials were small (n≤30) and brief (≤3 weeks). Most of the quetiapine and aripiprazole RCTs, in contrast, were large (n ≥100) and with a longer follow-up (6–12 weeks).
Duration of refractoriness did not influence outcomes, but staging of treatment resistance could not be examined as an outcome predictor due to insufficient variability on this measure.
As some augmentation treatments can have troublesome long-term AEs, future studies should address the issue of the duration needed to prevent relapse and recurrence. Lithium and T3/T4 need to be examined in adequately powered RCTs of sufficient duration (eg, 6–8 weeks). Brexpiprazole should be studied in independent trials for TRD, as it seems effective as adjunctive treatment.3
Yes, this study should influence clinical practice. First, the results discourage buspirone, bupropion, methylphenidate, lamotrigine and pindolol augmentation trials (no better than placebo), and olanzapine and risperidone augmentation trials (better efficacy results with aripiprazole and quetiapine). Next, the results favour aripiprazole or quetiapine augmentation for 6–12 weeks; however, the risk of discontinuation due to AEs is doubled to quadrupled, relative to placebo. So, a 3–4-week trial of lithium or T3/T4 seems the quickest and best option.
Competing interests: CA receives support from Sun Pharmaceuticals (money paid to charities), receives authorship royalties from Zydus Neurosciences, as well as grants from the Indian Council for Medical Research, the Department of Biotechnology, and the Department of Science and Technology.
Provenance and peer review: Commissioned; internally peer reviewed.
Malhi GS, Bassett D, Boyce P, et al. Royal Australian and New Zealand College of Psychiatrists clinical practice guidelines for mood disorders. Aust N Z J Psychiatry 2015; 49:1087–206. Mavridis D, Giannatsi M, Cipriani A, et al. A primer on network meta-analysis with emphasis on mental health. Evid Based Ment Health 2015; 18:40–6. McKeage K. Adjunctive brexpiprazole: a review in major depressive disorder. CNS Drugs 2016; 30:91–9. 
