Meta-analysis of noradrenergic and specific serotonergic antidepressant use in schizophrenia

Abstract

We examined whether noradrenergic and specific serotonergic antidepressants (NaSSAs: mirtazapine and mianserin), as augmentation therapy, have therapeutic potential for schizophrenia treatment. A systematic review was conducted of PubMed, Cochrane Library and PsycINFO in December 2012 and meta-analyses of double-blind, randomized placebo-controlled trials were performed. Standardized mean difference (SMD), risk ratio (RR), number-needed-to-treat (NNT), number-needed-to-harm (NNH) and 95% confidence intervals (CI) were calculated. Results were across 12 studies and 362 patients were included (mirtazapine: seven trials and 221 patients; mianserin: five trials and 141 patients). NaSSA augmentation therapy was superior to placebo in overall symptoms (s.m.d. = −0.75, CI −1.24 to −0.26, p = 0.003, N = 11, n = 301), negative symptoms (s.m.d. = −0.88, CI −1.41 to −0.34, p = 0.001, N = 9, n = 240) and response rate (RR = 0.71, CI 0.57–0.88, p = 0.002, NNT = 4, p<0.00001, N = 6, n = 163). There was no significant difference in positive symptoms, depressive symptoms or discontinuation rate between NaSSAs and placebo treatments. In addition, no patients who received NaSSAs developed worsening psychosis during the study. For individual NaSSAs, mirtazapine was superior to placebo in overall symptoms (s.m.d. = 0.98, CI = −1.74 to −0.22, p = 0.01, N = 7, n = 194), negative symptoms (s.m.d. = −1.25, CI −1.88 to −0.62, p = 0.0001, N = 6, n = 172) and response rate (RR = 0.70, p = 0.04, NNT = 4, p = 0.0004, N = 4, n = 119). Moreover, NaSSAs were associated with reduced akathisia score (p < 0.00001) and extrapyramidal symptom scales (p = 0.01). However, NaSSAs caused drowsiness/sedation/somnolence compared with placebo (RR = 3.52, p = 0.002, NNT = 6, p = 0.01, N = 8, n = 209). Our results indicate that NaSSA (especially mirtazapine) augmentation therapy improved overall and negative symptoms in patients with schizophrenia. Because the included studies were small, the results should be treated with caution.

Meta-analysis, mianserin, mirtazapine, NaSSAs, schizophrenia

Introduction

The negative symptoms of schizophrenia, including problems with motivation, social withdrawal, diminished affective responsiveness, speech and movement, contribute more to poor functional outcomes and quality of life for individuals with schizophrenia than do positive symptoms (Milev et al., 2005; Foussias and Remington, 2010). Moreover, the severity of negative symptoms is associated with cognitive dysfunctions (Kishi et al., 2010). However, available antipsychotic therapy is not relatively effective for treating negative symptoms and cognitive dysfunction of schizophrenia (Goldberg et al., 2007, 2010; Leucht et al., 2009a, b).

Recently, Singh et al. (2010) reported the meta-analysis of antidepressant augmentation therapy of efficacy in negative symptoms of schizophrenia. Fluoxetine and trazodone improved negative symptoms compared with placebo. However, mirtazapine and mianserin were not more efficacious on negative symptoms than placebo. Mirtazapine and mianserin are also called noradrenergic and specific serotonergic antidepressants (NaSSAs) and are differentiated from other antidepressants (Croom et al., 2009). Mirtazapine has a tetracyclic structure similar to that of mianserin. NaSSAs are an antagonist for noradrenaline (NA) α₂ autoreceptors, which results in enhanced release of NA from noradrenergic terminals, and for NA α₂ heteroreceptors (receptors on serotonin (5-HT) neurons), which increase serotonin release from serotonergic terminals. The increased release of NA also increases the firing of serotonergic neurons through stimulation of α₁ receptors. Both NA and 5-HT functions are increased. In addition, mirtazapine blocks post-synaptic 5-HT₂ and 5-HT₃ (but not 5-HT_1A) receptors; it is these actions that give rise to the description ‘specific serotonergic’. The net result is to increase NA and 5-HT transmission. Blockade of 5-HT₂ and 5-HT₃ receptors means that side-effects (insomnia, agitation, sexual dysfunction and nausea) resulting from stimulation of these receptors should be minimized. In addition, mirtazapine and mianserin also block histamine H₁ receptors very strongly, thus causing sedation, but have little effect on acetylcholine, dopamine, or NA α₁ receptors (Croom et al., 2009; Watanabe et al., 2011).

Although Singh et al. (2010) reported that mirtazapine and mianserin were not superior to placebo in negative symptoms (four studies: Berk et al., 2001, 2009; Zoccali et al., 2004; Joffe et al., 2009), recently, two randomized placebo-controlled trials (RCTs) of mirtazapine were reported (Abbasi et al., 2010; Cho et al., 2011). In addition, some studies reported that mirtazapine provided greater benefit for the improvement of overall symptoms, as well as negative symptoms, compared with placebo (Zoccali et al., 2004; Joffe et al., 2009; Abbasi et al., 2010). However, the results of other studies were inconsistent (Berk et al., 2001, 2009; Cho et al., 2011). Reasons for these discrepant results may be related to the small sample sizes of these trials, which had 9–20 participants in each treatment arm, and disparate outcome measures. A meta-analysis can increase the statistical power for group comparisons and can overcome the limitation of sample size in conditions where larger trials are lacking (Cohn and Becker, 2003). Moreover, using random effects models and standardized mean difference (s.m.d.) analyses, outcomes with different metrics can be combined (DerSimonian and Laird, 1986). Moreover, the study needed to describe the clinical pharmacological profile (efficacy: overall, positive, negative and depressive symptoms; tolerability: discontinuation; safety: each side-effect) of NaSSA augmentation therapy in patients with schizophrenia by conducting a systematic review and a meta-analysis. Therefore, we carried out a systematic review and meta-analysis of RCTs of NaSSA augmentation therapy in schizophrenia. Given the involvement of 5-HT and NA mechanisms in the beneficial effects on psychopathology, we hypothesized that NaSSA augmentation therapy might be superior to placebo in one or more efficacy outcomes.

Method

Search strategy, data extraction, and outcomes

To identify relevant studies, we searched PubMed, the Cochrane Library databases and PsycINFO citations in December 2012, using the following keywords: ‘aptazapine’ (Liebman et al., 1983), ‘esmirtazapine’ (Pinder and van Delft, 1983), ‘mirtazapine’, ‘mianserin’ or ‘setiptiline’ (Niho et al., 1986; Yamada and Furukawa, 1991) and ‘schizophrenia’. Inclusion criteria were double-blind, placebo-controlled, RCTs of NaSSA augmentation therapy. Complementing the electronic search, pertinent review articles, prior meta-analyses and reference lists from identified studies were hand-searched for additional studies eligible for inclusion in this meta-analysis. Two authors (T. K. and N. I.) double-checked the inclusion and exclusion criteria of the identified studies. When data required for the meta-analysis were missing, the first/corresponding authors were contacted for additional information. Additional, unpublished data were provided for this meta-analysis by Dr Seetal Dodd. Two authors (T. K. and N. I.) independently extracted data and checked and entered data into the Review Manager.

Data synthesis

We included the outcomes of this meta-analysis, which could include at least three studies in each outcome (Supplementary Table S1). The primary outcomes were the overall [total Brief Psychiatric Rating Scale (BPRS; Overall and Gorham, 1962] end-point scores from two studies, change of total BPRS scores from baseline to end-point from three studies, total Positive and Negative Syndrome Scale (PANSS; Kay et al., 1989) end-point score from three studies and change of total PANSS scores from baseline to end-point from three studies), positive [change of positive BPRS scores from baseline to end-point from one study, positive PANSS end-point score from four studies, change of positive PANSS scores from baseline to end-point from two studies, Scale for the Assessment of Positive Symptoms (SAPS; Andreasen, 1985) end-point score from two studies and change of SAPS scores from baseline to end-point from one study] and negative symptoms [negative PANSS end-point score from two studies, change of negative PANSS scores from baseline to end-point from two studies, Scale for the Assessment of Negative Symptoms (SANS; Andreasen, 1982) end-point score from three studies and change of SANS scores from baseline to end-point from two studies]. The secondary outcomes were PANSS general subscale score (change of PANSS general scores from baseline to end-point from two studies, PANSS general score from three studies), depressive symptoms [the Hamilton Rating Scale for Depression (HAMD; Hamilton, 1960) end-point score from three studies, change of negative HAMD scores from baseline to end-point from three studies, change of the Calgary Depression Rating Scale (CDSS; Addington et al., 1990) from baseline to end-point from one study and BPRS depression end-point score from one study], Clinical Global Impressions-Severity of Illness (CGI-S; Guy and Bonato, 1970) end-point score from two studies and change of score from baseline to end-point from one study, response rate (the definition of responder based on original study, ⩾20% reduction in SANS total score from two studies, ⩾30% reduction in SANS total score from one study, ⩾20% reduction in PANSS total score from one study, ⩾50% reduction in PANSS total score from one study and ⩾25% reduction in BPRS total score from one study) and discontinuation due to all-cause and side-effect. Moreover, we analysed reported adverse effects, such as drowsiness/sedation/somnolence, as long as ⩾3 studies contributed data to the analyses. For akathisia and extrapyramidal symptoms (EPS), we used Barnes Akathisia Rating Scale (BAS; Barnes, 1989) and Simpson–Angus Scale (SAS; Simpson and Angus, 1970), respectively.

Statistical analysis

Meta-analysis was conducted according to the guidelines from the Preferred Reporting Items for Systematic reviews and Meta-Analysis group (Moher et al., 2009). We based the analyses on intent-to-treat (ITT) or modified ITT data (i.e. at least one dose or at least one follow-up assessment). However, the data of completer analysis were not excluded to obtain as much information as possible (Poyurovsky et al., 1999; Berk et al., 2009). In addition, the Abbasi et al. (2010) study did not report actual numbers of patients who were analysed for all outcomes. However, they mentioned ‘ITT analysis with last observation carried forward procedure was performed and the sample size was calculated at least 15 in each group’. Therefore, we substituted n = 15 as the number of patients who were analysed. The meta-analysis was performed using Review Manager, Version 5.1 for Windows (Review Manager version 5.0, Cochrane Collaboration, http://ims.cochrane.org/revman). To combine studies, the random effects model by Der-Simonian and Laird (1986) was used in all cases. This method is more conservative, because it statistically accounts for the possibility that the meta-analysed studies and samples are heterogeneous. For continuous data, s.m.d. was used, combining effect size (Hedges’ g) data. For dichotomous data, pooled estimates of risk ratio (RR) were calculated together with 95% confidence intervals (CI). In the case of significant group differences in categorical variables, the number-needed-to-treat (NNT) or the number-needed-to-harm (NNH) for adverse effects was calculated. NNH was derived from the risk difference (RD) using the following formula: NNH = 1/RD, with the 95% CI of NNH being the inverse of the upper and lower limits of the 95% CI of the RD. We explored study heterogeneity using the χ² test of homogeneity (p < 0.05) together with I² statistics, considering values of ⩾50% to reflect considerable heterogeneity (Higgins et al., 2003). In cases of I² values >50% for the primary outcomes, we planned to conduct sensitivity analyses to determine the reasons for the heterogeneity. Finally, funnel plots were inspected visually to explore the possibility of publication bias.

Results

Search results and study, patient and treatment characteristics

The online search yielded 108 references using the above mentioned key words (mirtazapine and schizophrenia = 42 references, mianserin and schizophrenia = 65 references and setiptiline and schizophrenia = 1 reference). We excluded 90 studies based on the title and abstract review. Six studies were excluded based on full-text review because of same study. The online search identified 12 studies (Hayashi et al., 1997; Poyurovsky et al., 1999, 2003a, b; Berk et al., 2001, 2009; Shiloh et al., 2002; Zoccali et al., 2004; Miodownik et al., 2006; Joffe et al., 2009; Abbasi et al., 2010; Cho et al., 2011; Fig. 1). Although we conducted a hand-search using review articles, we did not identify any additional studies included in the meta-analysis. Twelve studies and 362 patients were included (mirtazapine: seven trials and 221 patients; mianserin: five trials and 141 patients; Table 1). The mean age of the study population was 38.1 yr. All studies were double-blind RCTs and published in English. Five of the 12 studies were conducted in Israel (Poyurovsky et al., 1999, 2003a, b; Shiloh et al., 2002; Miodownik et al., 2006) and no study was conducted in the USA. The mean study duration was 33.3 d (range 5–56 d). Sample sizes ranged from 18–40, with 9–20 participants being randomized to each of the groups. The study population was nine trials focused on only schizophrenia patients and three trials focused on schizophrenia/schizoaffective disorder/schizophreniform disorder. Antipsychotics studied in the RCTs included risperidone (two trials), haloperidol (one trial), clozapine (one trial), mixed second generation antipsychotic (SGA; one trial), mixed first generation antipsychotic (FGA; six trials) and mixed antipsychotic (one trial). All included studies were double-blind, randomized, placebo-controlled trials and stated the detailed information about study design. However, two studies used completer analysis (Supplementary Table S1).

Fig. 1

Preferred reporting items for systematic reviews and meta-analysis flow diagram.

Table 1

Study, patient and treatment characteristics of included double-blind randomized placebo-controlled trials

Study	Total (n)	Patients (%)	Diagnosis	Duration	Age (mean±s.d.)	Male (%)	Race (%)	Drug	n	Dose (dose mg/d)	Outcomes	Concomitant drugs (%)
Mirtazapine
Abbasi et al. (2010) (Iran) Non-industry	40	Chronic SZ (100). In-patients (100). Active phase of illness (PANSS total⩾60 and PANSS negative ⩾15). The patients did not receive AP for 1 wk prior to entering the trial or long acting APs at least 2 months before the study.	DSM-IV-TR	8 wk	Mir:33.35 ± 6.42, PBO: 34.50 ± 7.51	Mir: 65.0, PBO: 70.0	Iranian (100)	Mir + Ris	20	Mir:30 [fixed], Ris:6 [fixed]	PANSS total: Mir>PBO. PANSS positive: Mir = PBO, PANSS negative: Mir>PBO, PANSS general: Mir = PBO	Bip (NR), Overall mean Bip dosages (mg, mean±s.d.): Mir = 104.04 ± 109.215, PBO = 125.09 ± 88.28. no. of days of Bip use (d, mean±s.d.): Mir = 16.38 ± 17.54, PBO = 21.22 ± 12.48
								PBO + Ris	20	Ris:6 [fixed]
Berk et al. (2001) (South Africa) Non-industry	30	SZ (100). In-patients (100). All participants received Hal (5 mg/d)	DSM-IV and MINI	6 wk	29.5 ± 9.3	83.3	NR	Mir + Hal	15	Mir:30 [fixed], Hal:5 [fixed]	PANSS total: Mir>PBO. PANSS positive: Mir = PBO, PANSS negative: Mir>PBO, PANSS general: Mir = PBO, HAMD: Mir = PBO	Anti-CHO: Mir (40), PBO (86.7)
								PBO + Hal	15	Hal:5 [fixed]
Berk et al. (2009) (Australia) Industry	40	SZ (100). In-patients (NR) or out-patients (NR). All participants received SGA as usual [Clo (45), Que (21), Ris (18), Ola (9), Ari (6)]	MINI	6 wk	36.8 ± 9.8	85	NR	Mir + SGA	20	Mir:30 [fixed], SGA (Ris eq) = 7.10 ± 3.29 [fixed]	PANSS positive: Mir = PBO, PANSS negative: Mir = PBO, PANSS general: Mir = PBO, HAMD: Mir = PBO	Benz (NR)
								PBO + SGA	20	SGA (Ris eq) = 6.33 ± 3.54 [fixed]
Cho et al. (2011) (South Korea) Non-industry	21	SZ (100). Out-patients (100). All participants received Ris and CGI-S⩾4.	DSM-IV	8 wk	Mir:35.08 ± 13.58, PBO: 36.44 ± 9.57	Mir: 45.4, PBO: 55.6	Korean (100%)	Mir + Ris	12	Mir:30 [fixed], Ris: 3 ± 1.94 [fixed]	PANSS total: Mir = PBO. PANSS positive: Mir = PBO, PANSS negative: Mir>PBO, PANSS general: Mir = PBO	BTP: Mir (55), PBO (78), Benz: Mir (67), PBO (87), Pro (NR)
								PBO + Ris	9	Ris: 4.22 ± 1.83 [fixed]
Mirtazapine
Joffe et al. (2009) (Finland) Non-industry	40	SZ (97.5) and SA with depressive type. CGI-S⩾4 (⩾6 previous weeks). The patients had to receive ⩾1 FGAs [⩾400 mg CHL eq, unchanged for ⩾6 previous weeks (8 wk for depot FGAs)]. In-patients (Mir = 55%, PBO = 36.8). Number of patients: Hal = 14, Tri = 13, Hal-D = 10, Flu-D = 8, Lev = 6, CPT = 5, Zuc-D = 5, Zuc = 4, FLP = 2, PRZ = 2, Sul = 1.	DSM-IV-TR	6 wk	Mir:43.40 ± 9.24, PBO: 48.21 ± 9.68	Mir: 55.0, PBO: 47.4	NR	Mir + FGA	20	Mir:30, FGA (CHL eq) = 330.75 ± 123.53 [fixed]	PANSS total: Mir = PBO. PANSS positive: Mir>PBO, PANSS negative: Mir>PBO, PANSS general: Mir = PBO, CDSS: Mir = PBO	Benz (NR)
								PBO + FGA	20	FGA (CHL eq) = 316.58 ± 164.87 [fixed]
Poyurovsky (2003a) (Israel) Non-industry	26	SZ (100). In-patients (100). All participants had mild akathisia (BAS⩾2), and received FGAs (Hal = 5–15 mg/d, Per = 8–24 mg/d).	DSM-IV	5 d	Mir: 31.8 ± 10.9, PBO: 29.1 ± 9.9	Mir: 38.5, PBO: 61.5	NR	Mir + FGA	13	Mir:15, FGA: [fixed]	PANSS: Mir = PBO, HAMD: Mir = PBO	NR
								PBO + FGA	13	FGA: [fixed]
Zoccali et al. (2004) (Italy) Non-industry	24	SZ. All patients had demonstrated persistent negative symptoms despite an adequate trial of Clo (Clo 150–650 mg/d for at least 1 yr)	DSM-IV	8 wk	Mir:30.7 ± 6.5, PBO: 33.4 ± 9.0	Mir: 50.0, PBO: 70.0	NR	Mir + Clo	12	Mir:30 [fixed], Clo = 320 ± 151.2	BPRS: Mir>PBO, SAPS: Mir = PBO, SANS: Mir>PBO, BPRS depression: Mir = PBO	NR
								PBO + Clo	12	Clo = 325 ± 131.7
Mianserin
Hayashi et al. (1997) (Japan) Non-industry	26	SZ. Age⩾50 yr and duration of illness ⩾10 yr. Patients had moderate to severe negative symptoms and tardive dyskinesia and received FGA.	DSM-III-TR	5 wk	Mia:64.6 ± 7.8, PBO: 59.9 ± 9.3	Mia: 61.5, PBO: 53.8	Japanese (100)	Mia + FGA	13	Mia: 20–60 (15.8 + /−6.1) [flexible], FGA(CHL eq) = 492 ± 280 [fixed]	BPRS: Mia = PBO, BPRS positive: Mia = PBO, SANS: Mia>PBO	NR
								PBO + FGA	13	FGA (CHL eq) = 740 ± 531 [fixed]
Mianserin
Miodownik et al. (2006) (Israel) Non-industry	37	SZ (67.6) and SA. In-patients (100). All participants had mild akathisia (BAS⩾2)	DSM-IV	5 d	Mia:39.1 ± 13.0, PBO: 43.8 ± 13.0	Mia: 60.0, PBO: 52.9	NR	Mia + AP	20	Mia: 15 [fixed], SGA (40%) [fixed], FGA (60%) [fixed]	BPRS: Mia>PBO	TPH: Mia (20), PBO (17.6)
								PBO + AP	17	SGA(35.3%) [fixed], FGA (64.7%) [fixed]
Poyurovsky et al. (1999) (Israel) Non-industry	30	SZ (53.3%), SA (16.7%) and SF (30%). All participants were in-patients, had mild akathisia (BAS⩾2), and received FGA (Hal, Per, Flu)	DSM-IV	5 d	Mia:30.1 ± 8.5, PBO: 29.3 ± 10.0	Mia: 66.7, PBO: 53.3	NR	Mia + FGA	15	Mia: 15 [fixed], FGA [fixed]	BPRS: Mia>PBO, HAMD: Mia = PBO	Anti-CHO (TPH and Bip): Mir (46.7), PBO (46.7), Benz: Mir (53.3), PBO (66.7),
								PBO + FGA	15	FGA [fixed]
Poyurovsky (2003b) (Israel) Non-industry	30	SZ (100). In-patients (100)(12–36 months). All participants received FGA (Hal (50), CHL (40), Per (10)).	DSM-IV	4 wk	Mia:42.5 ± 12.9, PBO: 45.5 ± 7.5	70	NR	Mia + FGA	15	Mia: 15 [fixed], FGA [fixed]	SAPS: Mia = PBO, SANS: Mia = PBO, HAMD: Mia = PBO	Anti-CHO (TPH and Bip): Mir (93.3), PBO (93.3), Benz: Mir (66.7), PBO (53.3), VAL: Mir (20), PBO (13.3),
								PBO + FGA	15	FGA [fixed]
Shiloh (2002) (Israel) Non-industry	18	Treatment-resistant SZ (100). In-patients (100). Failed response at least 3 AP, and received FGA (Hal (50), Per (50).	DSM-IV	6 wk	Mia:32.8 ± 9.0, PBO: 42.6 ± 7.3	Mia: 55.6, PBO: 66.7	NR	Mia + FGA	9	Mia: 30 [fixed], FGA (CHL eq) = 444.4 ± 170.4 [fixed]	BPRS: Mia>PBO, SAPS: Mia = PBO, SANS: Mia = PBO, HAMD: Mia = PBO	THP: Mir (55.6), PBO (55.6), Pro: only PBO (22.2)
								PBO + FGA	9	FGA (CHL eq) = 566.7 ± 229.6 [fixed]

Anti-CHO, Anticholinergic drugs; AP, antipsychotic; Ari, aripiprazole; BAS, Barnes Akathisia Rating Scale; Benz, benzodiazepines; BTP, benztropine; Bip, biperiden; BPRS, Brief Psychiatric Rating Scale; CGI-S, Clinical Global Impression of Severity; CDSS, Calgary Depression Scale for Schizophrenia; CPT, chlorprotixene; CHL eq, chlorpromazine equivalent; Clo, clozapine; DSM, Diagnostic and Statistical Manual of Mental Disorders; FGA, first generation antipsychotic; FLP, flupentixol; Flu, fluphenazine; Flu-D, fluphenazine decanoate; Hal, haloperidol; Hal-D, haloperidol decanoate; HAMD, Hamilton Rating Scale for Depression; Lev, levomepromazine; MINI, Mini-International Neuropsychiatric Interview; Mia, mianserin; Mir, mirtazapine; MD, mood stabilizer; NR, not reported; Ola, olanzapine; PANSS, Positive and Negative Syndrome Scale; PRZ, periciazine; Per, perphenazine; PBO, placebo; Pro, propranolol; Que, quetiapine; Ris, risperidone; SA, schizoaffective disorder; SAPS, Scales for the Assessment of Positive Symptoms; SANS, Scale for the Assessment of Negative Symptoms; SF, schizophreniform disorder; SGA, second generation antipsychotic; SZ, schizophrenia; Sul, sulpiride; Tri, trifluoperazine; THP, trihexyphenidyl; VAL, valproic acid; Zuc, zuclopentixol; Zuc-D, zuclopentixol decanoate.

Results of efficacy

NaSSA augmentation therapy was superior to placebo in overall symptoms (s.m.d. = −0.75, CI −1.24, −0.26, p = 0.003, I² = 74, N = 11, n = 301; Supplementary Fig. S1), negative symptoms (s.m.d. = −0.88, CI −1.41 to −0.34, p = 0.001, I² = 72, N = 9, n = 240; Supplementary Fig. S2) and response rate (RR = 0.71, CI 0.57–0.88, p = 0.002, NNT = 4, p < 0.00001, N = 6, n = 163; Table 2). NaSSA augmentation therapy was also better than placebo for providing marginal improvement of depressive symptoms (s.m.d. = −0.24, p = 0.09, N = 8, n = 207). However, NaSSA augmentation therapy was not superior to placebo in positive symptoms (s.m.d. = −0.28, p = 0.18, N = 9, n = 240), CGI-S (s.m.d. = −0.65, p = 0.13, N = 3, n = 100) or PANSS general score (s.m.d. = −0.31, p = 0.22, N = 5, n = 152; Table 2). The data in each efficacy outcome were simulated with no publication bias.

Table 2

NaSSAs vs. placebo, efficacy and discontinuation

Outcome	NaSSAs	N	n	I2	s.m.d.	95% CI	p
Overall symptoms	Mirtazapine	7	194	82	−0.98	−1.74 to −0.22	0.01
	Mianserin	4	107	26	−0.46	−0.92 to 0.00	0.05
	Combined	11	301	74	−0.75	−1.24 to −0.26	0.003
Positive symptoms	Mirtazapine	6	172	70	−0.24	−0.82 to 0.33	0.40
	Mianserin	3	68	8	−0.34	−0.84 to 0.17	0.19
	Combined	9	240	58	−0.28	−0.69 to 0.13	0.18
Negative symptoms	Mirtazapine	6	172	71	−1.25	−1.88 to −0.62	0.0001
	Mianserin	3	68	0	−0.15	−0.63 to 0.33	0.54
	Combined	9	240	72	−0.88	−1.41 to −0.34	0.001
PANSS general subscale scores	Mirtazapine	5	152	56	−0.31	−0.81 to 0.19	0.22
Depressive symptoms	Mirtazapine	5	139	29	−0.35	−0.76 to 0.06	0.09
	Mianserin	3	68	0	−0.04	−0.52 to 0.44	0.86
	Combined	8	207	2	−0.24	−0.52 to 0.04	0.09
Clinical Global Impression-Severity	Mirtazapine	1	39	NA	−0.63	−1.28 to 0.01	0.05
	Mianserin	2	61	88	−0.65	−2.21 to 0.91	0.42
	Combined	3	100	76	−0.65	−1.50 to 0.20	0.13
	NaSSAs	N	n	I2	RR	95% CI	p
Response rate for psychopathology	Mirtazapine	4	119	57	0.70	0.49–0.99	0.04
	Mianserin	2	44	0	0.66	0.49–0.91	0.01
	Combined	6	163	33	0.71	0.57–0.88	0.002
Discontinuation due to all cause	Mirtazapine	7	221	0	1.24	0.57–2.71	0.59
	Mianserin	5	141	71	0.59	0.03–11.32	0.73
	Combined	12	362	0	1.19	0.60–2.33	0.62
Discontinuation due to side-effect	Mirtazapine	7	221	0	1.11	0.34–3.58	0.87
	Mianserin	4	111	NA	3.00	0.13–68.26	0.49
	Combined	11	332	0	1.25	0.42–3.76	0.69

NaSSAs, Noradrenergic and specific serotonergic antidepressants; N, number of studies; n, number of patients; s.m.d., standardized mean difference; CI, confidence intervals; PANSS, Positive and Negative Syndrome Scale; NA, not applicable; RR, risk ratio.

p values < 0.05 are shown in bold.

For individual NaSSAs, mirtazapine was superior to placebo in overall symptoms (s.m.d. = −0.98, CI −1.74 to −0.22, p = 0.01, N = 7, n = 194), negative symptoms (s.m.d. = −1.25, CI −1.88 to −0.62, p = 0.0001, N = 6, n = 172) and response rate (RR = 0.70, p = 0.04, NNT = 4, p = 0.0004, N = 4, n = 119; Table 2). Mirtazapine was also better than placebo for providing marginal improvement of depressive symptoms (s.m.d. = −0.35, p = 0.09, N = 5, n = 139) and CGI-S scores (s.m.d. = −0.63, p = 0.05, N = 1, n = 39). Mianserin was superior to placebo in only response rate (RR = 0.66, p = 0.01, NNT = 3, p = 0.0007, N = 2, n = 44; Table 2).

We found significant heterogeneity in overall, positive and negative symptoms between NaSSAs and the placebo treatment group. However, there was no significant heterogeneity in these outcomes in mianserin and placebo. Therefore, we performed the sensitivity analysis of only meta-analysis of mirtazapine to seek out confounding factors. When excluding the data of completer analysis (Berk et al., 2009) and clozapine trial (Zoccali et al., 2004) in overall symptoms, the significant heterogeneity disappeared (s.m.d. = −0.84, CI −1.19 to −0.49, p < 0.00001, I² = 0, N = 5, n = 139). Also, when excluding the Joffe et al. (2009) study in positive symptoms, the significant heterogeneity disappeared (s.m.d. = −0.02, CI −0.51 to −0.46, p = 0.92, I² = 48, N = 5, n = 133). In addition, when excluding the data of completer analysis (Berk et al., 2009), significant heterogeneity was not found (s.m.d. = −1.48, CI −1.86 to −1.09, p < 0.00001, I² = 0%, N = 5, n = 139). However, the results of these outcomes did not change. While mirtazapine augmentation therapy in patients primarily treated with SGAs was significantly superior to placebo for negative symptoms (s.m.d. = −1.17, CI −2.14 to −0.20, p = 0.02, I² = 79, N = 4, n = 103), mirtazapine augmentation therapy in patients primarily treated with SGAs was marginally superior to placebo for overall symptoms (s.m.d. = −1.31, CI −2.82 to −0.20, p = 0.09, I² = 90, N = 4, n = 105) and response rate (RR = 0.59, p = 0.07, I² = 60, N = 3, n = 80). Conversely, mirtazapine augmentation therapy improved both overall (s.m.d. = −0.90, CI −1.34 to −0.46, p < 0.0001, I² = 0, N = 3, n = 89) and negative symptoms (s.m.d. = −1.24, CI −1.94 to −0.87, p < 0.00001, I² = 0, N = 2, n = 69) compared with placebo groups in patients primarily treated with FGAs.

Results of tolerability and side-effects

There were no significant differences in discontinuation rate between NaSSAs and placebo treatments (discontinuation for the following reasons; all-cause: RR = 1.19, CI 0.60–2.33, p = 0.62, I² = 0%, N = 12, n = 362, side-effects: RR = 1.25, CI 0.42–3.76, p = 0.69, I² = 0%, N = 11, n = 332; Table 2). In addition, it was reported that there were no patients who received NaSSAs who developed worsening psychosis during the study and dropped out from the study because of inefficacy or relapse.

Although we conducted meta-analyses on seven side-effects (at least one side-effect, dizziness/orthostatic hypotension, drowsiness/sedation/somnolence, headache, psychosis exacerbation/worsening of schizophrenia, tremor/Parkinsonism), we found only significant differences in drowsiness/sedation/somnolence between NaSSAs and the placebo treatment group (RR = 3.52, CI 1.61–7.71, p = 0.002, I² = 0%, NNH = 6, p = 0.01, N = 8, n = 209; Table 3; Supplementary Fig. S3). For individual NaSSAs, both mirtazapine and mianserin were associated with higher risk drowsiness/sedation/somnolence compared with placebo (mirtazapine: RR = 5.46, p = 0.02, NNH = 5, p = 0.0003 N = 3, n = 86; mianserin: RR = 2.91, p = 0.03, NNH = n.s., N = 5, n = 123; Table 3; Supplementary Fig. S3).

Table 3

NaSSAs vs. placebo, side-effect

Side-effect	NaSSAs	N	n	I2	RR	95% CI	p
At least one side-effect	Mirtazapine	2	79	0	1.00	0.91–1.10	0.98
	Mianserin	1	37	NA	NA
	Combined	3	116	0	1.00	0.91–1.10	0.98
Dizziness/orthostatic hypotension	Mirtazapine	1	40	NA	1.67	0.46–6.05	0.44
	Mianserin	3	93	0	3.00	0.33–27.23	0.33
	Combined	4	133	0	1.94	0.64–5.90	0.24
Drowsiness/sedation/somnolence	Mirtazapine	3	86	0	5.46	1.31–22.75	0.02
	Mianserin	5	123	0	2.91	1.14–7.44	0.03
	Combined	8	209	0	3.52	1.61–7.71	0.002
Headache	Mirtazapine	3	100	0	1.48	0.51–4.33	0.47
	Mianserin	1	37	NA	NA
	Combined	4	137	0	1.48	0.51–4.33	0.47
Psychosis exacerbation/worsening of schizophrenia	Mianserin	3	93	NA	NA
Tremor/Parkinsonism	Mirtazapine	2	70	0	0.64	0.33–1.25	0.19
	Mianserin	2	55	NA	1.00	0.44–2.29	1.00
	Combined	4	125	0	0.76	0.45–1.28	0.31

NaSSAs, Noradrenergic and specific serotonergic antidepressants; N, number of studies; n, number of patients; NA, not applicable; s.m.d., standardized mean difference; CI, confidence intervals; RR, risk ratio.

p values < 0.05 are shown in bold.

Regarding akathisia and EPS, the patients who received NaSSAs had lower SAS and BAS scores than patients who received placebo (BAS-objective: s.m.d. = −1.19, p < 0.00001; BAS-distress: s.m.d. = −1.22, p < 0.00001; BAS-global: s.m.d. = −1.48, p < 0.00001; SAS: s.m.d. = −0.56, p = 0.01; Table 4).

Table 4

NaSSAs vs. placebo, akathisia and extrapyramidal symptoms

Outcome	NaSSAs	N	n	I2	s.m.d.	95% CI	p
BAS-objective	Mirtazapine	1	20	NA	−1.44	−2.44 to −0.43	0.005
	Mianserin	2	63	0	−1.12	−1.65 to −0.58	<0.0001
	Combined	3	83	0	−1.19	−1.67 to −0.71	<0.00001
BAS-subjective	Mirtazapine	1	20	NA	−0.71	−1.62 to 0.20	0.12
	Mianserin	2	63	87	−0.83	−2.30 to 0.65	0.27
	Combined	3	83	73	−0.79	−1.70 to 0.11	0.09
BAS-distress	Mirtazapine	1	20	NA	−0.75	−1.68 to 0.15	0.10
	Mianserin	2	63	0	−1.39	−1.95 to −0.83	<0.00001
	Combined	3	83	0	−1.22	−1.70 to −0.74	<0.00001
BAS-global	Mirtazapine	1	20	NA	−1.48	−2.50 to −0.47	0.004
	Mianserin	2	63	0	−1.48	−2.05 to −0.91	<0.00001
	Combined	3	83	0	−1.48	−1.98 to −0.98	<0.00001
SAS	Mirtazapine	1	36	NA	−0.50	−1.17 to 0.16	0.14
	Mianserin	2	50	0	−0.60	−1.18 to −0.02	0.04
	Combined	3	86	0	−0.56	−1.00 to −0.12	0.01
	NaSSAs	N	n	I2	RR	95% CI	P
Response rate for akathisia	Mirtazapine	1	26	NA	0.50	0.27–0.92	0.03
	Mianserin	2	67	56	0.51	0.29–0.89	0.02
	Combined	3	93	11	0.52	0.38–0.72	<0.0001

BAS, Barnes Akathisia Scale; SAS, Simpson–Angus Scale; N, number of studies; n, number of patients; NA, not applicable; s.m.d., standardized mean difference; CI, confidence intervals; RR, risk ratio.

p values < 0.05 are shown in bold.

Discussion

This is the first systematic review and meta-analysis to focus on the efficacy and tolerability of NaSSA augmentation therapy in patients with schizophrenia. While NaSSA augmentation therapy did not differ from placebo regarding positive symptoms and PANSS general subscale scores, NaSSAs (especially mirtazapine) augmentation therapy significantly exhibited better overall and negative symptoms than placebo with large effect size (s.m.d.: overall symptoms = −0.75, negative symptoms = −0.88) and were also well tolerated compared with placebo (no significant differences in discontinuation due to all-cause as well as side-effect). We also detected marginal differences in depressive symptoms (p = 0.09) between NaSSA augmentation therapy (especially mirtazapine) and placebo. Because the included studies in the meta-analysis were small, the non-significant results could be because of low statistical power (i.e. type II error). NaSSA augmentation therapy did not produce worsening of schizophrenia compared with placebo in all included studies in the meta-analysis. From the above results, NaSSA augmentation therapy may be beneficial for the treatment of schizophrenia with regard to overall as well as negative symptoms.

Many reports insist that SGAs show efficacy in treating negative symptoms and that this is an important difference between SGAs and FGAs. However, Leucht et al. (2009a) reported that, although amisulpride, clozapine, olanzapine and risperidone showed greater efficacy than FGAs in the treatment of negative symptoms, aripiprazole, quetiapine, sertindole, ziprasidone and zotepine were not superior to FGAs in the treatment of negative symptoms. However, since clozapine and olanzapine have a high risk of cardio-metabolic disorder (Correll et al., 2009, 2011), clinicians have to use these antipsychotics with caution. Risperidone is also reported to have a risk for obesity (Correll et al., 2009, 2011) and hyperprolactinemia (Kishi et al., 2012; Moteshafi and Stip, 2012). Recently, it has been recommended that the antipsychotics that have less of a risk of side-effects, such as aripiprazole, should be selected for the treatment of schizophrenia (Buchanan et al., 2010). However, aripiprazole does not benefit the treatment of negative symptoms compared with FGAs (Leucht et al., 2009a). The results of our meta-analysis may suggest that NaSSAs (especially mirtazapine) enhance the efficacy (especially negative symptoms) of aripiprazole.

The EPS and akathisia also lead to the discontinuation of antipsychotic treatment of schizophrenia (Kane et al., 2009, 2010; Rummel-Kluge et al., 2012). Risperidone is also reported to be associated with more use of anti-Parkinson medication than clozapine, olanzapine, quetiapine and ziprasidone and has a risk of EPS (Rummel-Kluge et al., 2012). Aripiprazole is also reported to produce more akathisia than olanzapine and was associated more with the use of anti-Parkinson medication than olanzapine (Rummel-Kluge et al., 2012). On the other hand, while blonanserin, which is available for the treatment of schizophrenia only in Japan and Korea, is more efficacious for negative symptoms than FGA, and has lower risk of cardio-metabolic risk, it also produces more akathisia than risperidone (Kishi et al., 2012). Although EPS data were limited, and most studies used anticholinergic drugs in the current meta-analysis, our results show that NaSSAs seem to benefit the treatment of akathisia and EPS compared with placebo. The results of our meta-analysis may suggest that NaSSAs cover these deficits (akathisia and EPS) of aripiprazole and blonanserin. Larger trials are needed to extend these findings and to determine if NaSSAs influence the success of aripiprazole or blonanserin treatment in schizophrenia.

Since there were few studies that reported the efficacy of NaSSA augmentation therapy for cognitive function in schizophrenia patients, we were unable to perform a meta-analysis. However, Cho et al. (2011) reported that mirtazapine augmentation was superior to placebo management in ameliorating some cognitive impairment (vocabulary test and immediate memory). Stenberg et al. (2010) also reported that mirtazapine augmentation was superior to placebo in areas of visual-spatial ability and general mental speed/attention control. Moreover, not only speed but also accuracy of performance improved in patients with mianserin augmentation, when compared with the placebo group (Poyurovsky et al., 2003b). Thus, these results suggest that NaSSAs have cognitive-enhancing effects in patients with schizophrenia.

The results of this study need to be interpreted within its limitations. The main limitation is the small number of studies included. Future investigations of the therapeutic potential of NaSSAs should include a larger sample size. In particular, there were only four RCTs of SGAs included in the meta-analysis (two risperidone trials, one clozapine trial, one mixed SGA trial). It remains unclear which antipsychotic should be used in association with NaSSA. The second limitation is that patients with schizophrenia have been reported to have poor adherence to medication regimens, making it difficult to firmly establish the effectiveness of pharmacological interventions in this population (Lopez-Munoz and Rubio, 2011). Finally, two of the 12 trials included in this meta-analysis were short in duration (5 d). Future research should investigate long-term efficacy and generate more safety data.

In conclusion, our results indicate that NaSSA (especially mirtazapine) augmentation therapy improves overall and negative symptoms in patients with schizophrenia, and is well tolerated. Although EPS data were limited, our results show that NaSSAs seem to benefit the treatment of akathisia and EPS compared with placebo. Because the included studies were small, results should be treated with caution.

Acknowledgement

We thank Seetal Dodd, MD (The University of Melbourne) for providing additional information for the study.

Statement of Interest

Dr Kishi has received speaker's honoraria from Abbott, Astellas, Daiichi Sankyo, Dainippon Sumitomo, Eli Lilly, Yoshitomi, Otsuka, Meiji, Shionogi, Tanabe-Mitsubishi, Novartis and Pfizer. Dr Iwata has received speaker's honoraria from Astellas, Dainippon Sumitomo, Eli Lilly, GlaxoSmithKline, Janssen, Yoshitomi, Otsuka, Meiji, Shionogi, Novartis and Pfizer.

Supplementary material

Supplementary material accompanies this paper on the Journal's website.

Supplementary Material

Supplementary information supplied by authors.

Supplementary Material

Supplementary information supplied by authors.

Supplementary Material

Supplementary information supplied by authors.

Supplementary Material

Supplementary information supplied by authors.