Impact of sedative and appetite-increasing properties on the apparent antidepressant efficacy of mirtazapine, selective serotonin reuptake inhibitors and amitriptyline: an item-based, patient-level meta-analysis

Abstract

Background: In an influential network meta-analysis, the tricyclic antidepressant (TCA) amitriptyline was found to be the most efficacious of 21 antidepressants, hence outranking selective serotonin reuptake inhibitors (SSRIs) and serotonin and noradrenaline reuptake inhibitors (SNRIs). The alpha₂/5HT_2A/_2C/₃/H₁ antagonist mirtazapine was ranked as the second most effective and appeared at least as effective as the SSRIs and SNRIs that followed next. Since the most common effect parameter in depression trials-the sum score of the Hamilton Depression Rating Scale (HDRS-17-sum)-includes three items measuring sleep and two measuring appetite and weight, this outcome could be the result of amitriptyline and mirtazapine being more sedative and orexigenic. The main aim of this study was to compare mirtazapine with SSRIs or amitriptyline with respect to impact on core depression symptoms.

Methods: Access to patient-level data from all company-sponsored, acute-phase, HDRS-based, and randomized trials of mirtazapine in adult major depression available to Merck was granted. Thirty-two studies compared mirtazapine to placebo and/or amitriptyline or an SSRI whereas five compared mirtazapine to another TCA or an SNRI, venlafaxine. Data were divided into subgroups for direct comparisons of mirtazapine vs placebo or different subgroups of antidepressants. Indirect comparisons of SSRIs vs amitriptyline were also undertaken. Mixed models for repeated measures were used to assess efficacy as reflected by i) HDRS-17-sum, ii) six core depression symptoms (HDRS-6-sum), and iii) all individual items.

Findings: The dataset consisted of 5974 participants. Mirtazapine (n = 1362) outperformed SSRIs (n = 1369) on HDRS-17-sum, but this was due to differences regarding items reflecting sleep, appetite, and gastrointestinal dysfunction-with respect to reducing depressed mood, suicidality, and psychic anxiety, SSRIs and/or venlafaxine were more effective. Amitriptyline (n = 622) was superior to mirtazapine (n = 606) in reducing depressed mood, and the combined group of all TCAs (n = 831) outperformed mirtazapine (n = 824) also with respect to other core depression symptoms. Since there were no head-to-head comparisons of amitriptyline vs SSRIs, no firm conclusion may be drawn with respect to relative efficacy of the two, but indirect comparisons support the notion that amitriptyline and other tricyclics may be superior also to SSRIs.

Interpretation: While the apparent superiority of mirtazapine over SSRIs is explained by its sedative and orexigenic properties, and by its absence of gastrointestinal side effects, amitriptyline appeared more effective in reducing core symptoms of depression than mirtazapine and possibly also than SSRIs; given the indirect nature of the latter comparison, this outcome should however be interpreted with caution. Lack of information regarding dosing was another important limitation. The study illustrates the value of item-based analyses when assessing the relative efficacy of antidepressants.

Funding: The Swedish Research Council, the Swedish Brain Foundation. The Gothenburg Society of Medicine, the Swedish Society of Medicine, Åke Wiberg's Foundation, Märta Lundqvist's Foundation, Fredrik and Ingrid Thuring's Foundation, Söderström-Königska Foundation and Frimurare-Barnhusdirektionen in Gothenburg.

Keywords: Amitriptyline; Depression; Meta-analysis; Mirtazapine; SNRIs; SSRIs.

Fig. 1

Effect sizes over time for mirtazapine vs placebo, mirtazapine vs SSRIs, mirtazapine vs venlafaxine, and mirtazapine vs amitriptyline with respect to HDRS-17-sum, HDRS-6-sum and depressed mood. Legend: Shown are effect sizes with 95% confidence intervals for HDRS-17-sum, HDRS-6-sum, and the depressed mood item when comparing mirtazapine with placebo, SSRIs, venlafaxine, and amitriptyline, respectively. A positive effect size suggests superiority of mirtazapine. A negative effect size suggests superiority of the comparator. If a confidence interval does not cross zero, the difference is statistically significant. a: Effect size for HDRS-17-sum, b: Effect size for HDRS-6-sum, c: Effect size for depressed mood.

Fig. 2

Item-wise separation for mirtazapine vs placebo ager one and six weeks of treatment. Legend: The green centre line represents no difference between mirtazapine and placebo. Bars pointing towards the outer circle favour mirtazapine, where a bar reaching the outer circle would correspond to a mean difference (MD) of +0.4 points. Bars pointing towards the inner circle favour placebo where a bar reaching the inner circle would correspond to a mean difference of −0.4 points. ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001, ∗∗∗∗ = p < 0.0001.

Fig. 3

Item-wise separation for mirtazapine vs SSRIs, mirtazapine vs venlafaxine and mirtazapine vs amitriptyline, ager one and six weeks of treatment. Legend: The green centre line represents no difference between mirtazapine and control. Bars pointing towards the outer circle favour mirtazapine, where a bar reaching the outer circle would correspond to a mean difference (MD) of +0.4 points. Bars pointing towards the inner circle favour the active comparator where a bar reaching the inner circle would correspond to a mean difference of −0.4 points. ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001, ∗∗∗∗ = p < 0.0001.

Fig. 4

Item-wise separation for amitriptyline vs SSRIs and amitriptyline vs venlafaxine ager two and six weeks of treatment. Legend: The green centre line represents no difference between amitriptyline and the SSRIs and venlafaxine, respectively. Bars pointing towards the outer circle favour amitriptyline, where a bar reaching the outer circle would correspond to a mean difference (MD) of +0.4 points. Bars pointing towards the inner circle favour SSRIs and venlafaxine, respectively, where a bar reaching the inner circle would correspond to a mean difference (MD) of +0.4 points. ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001, ∗∗∗∗ = p < 0.0001.