Introduction
Schizophrenia is a heterogeneous clinical syndrome that includes many domains of psychopathology, with patients varying in their pathological manifestations (Buchanan & Carpenter, Reference Buchanan and Carpenter1994). People with schizophrenia experience ‘positive’ psychotic symptoms, including delusions, hallucinations, and disorganized thoughts and behavior, as required by diagnostic criteria. Research on the clinical manifestations of negative symptoms indicates that there are at least five symptoms, including blunted affect, alogia, avolition, social withdrawal, and anhedonia. Factor analysis suggests that these characteristics can be divided into two subdimensions: reduced expression and avolition/apatheia (Blanchard & Cohen, Reference Blanchard and Cohen2006; Correll & Schooler, Reference Correll and Schooler2020; Foussias et al., Reference Foussias, Agid, Fervaha and Remington2014; Kirkpatrick et al., Reference Kirkpatrick, Fenton, Carpenter and Marder2006, Reference Kirkpatrick, Strauss, Nguyen, Fischer, Daniel, Cienfuegos and Marder2011; Stiekema et al., Reference Stiekema, Liemburg, van der Meer, Castelein, Stewart, van Weeghel and Bruggeman2016). The presence of negative symptoms and cognitive impairment is frequently observed in schizophrenia and is known to be a significant contributor to impaired functional abilities (Villalta-Gil et al., Reference Villalta-Gil, Vilaplana, Ochoa, Haro, Dolz, Usall and Cervilla2006). Negative symptoms are the primary etiological factor contributing to compromised occupational (Velligan et al., Reference Velligan, Alphs, Lancaster, Morlock and Mintz2009; Ventura et al., Reference Ventura, Hellemann, Thames, Koellner and Nuechterlein2009) and social functioning (Lin et al., Reference Lin, Huang, Chang, Chen, Lin, Tsai and Lane2013), as well as imposing a significant burden on caregivers of individuals affected (Magliano et al., Reference Magliano, Fadden, Madianos, de Almeida, Held, Guarneri and Maj1998, Reference Magliano, Fadden, Economou, Held, Xavier, Guarneri and Maj2000). Consequently, negative symptoms play a pivotal role in the considerable financial implications of schizophrenia for healthcare systems and society (McEvoy, Reference McEvoy2007) and have been recognized as an unaddressed therapeutic requirement (Kirkpatrick et al., Reference Kirkpatrick, Fenton, Carpenter and Marder2006).
Despite considerable improvements in our understanding of schizophrenia and its multifaceted symptoms involving thinking, perception, and affect, drugs that effectively treat all of these aspects are still lacking (Kane & Correll, Reference Kane and Correll2010; Maric et al., Reference Maric, Jovicic, Mihaljevic and Miljevic2016). Most of the currently approved antipsychotic medications exert their effects through dopamine regulation (Carlsson & Carlsson, Reference Carlsson and Carlsson2006), primarily used to treat positive symptoms, agitation, and hostility. However, important functional domains of negative symptoms and cognitive dysfunction remain relatively unaddressed (Carbon & Correll, Reference Carbon and Correll2014). This fact has stimulated research beyond dopamine neurotransmitter regulation, aiming to address not only positive symptoms but also other areas of schizophrenia (Köster, Carbon, & Correll, Reference Köster, Carbon and Correll2014).
The treatment of negative symptoms has also evolved from a single antipsychotic drug to a combination of other treatments, such as physical intervention (Fitzgerald & Daskalakis, Reference Fitzgerald and Daskalakis2008; Freitas et al., Reference Freitas, Fregni and Pascual-Leone2009), psychological interventions (Mössler et al., Reference Mössler, Chen, Heldal and Gold2011; Sarin et al., Reference Sarin, Wallin and Widerlöv2011; Wykes et al., Reference Wykes, Steel, Everitt and Tarrier2008), anti-Parkinson drugs (Mössler et al., Reference Mössler, Chen, Heldal and Gold2011; Paraschakis, Reference Paraschakis2014; Rezaei et al., Reference Rezaei, Mohammad-Karimi, Seddighi, Modabbernia, Ashrafi, Salehi and Akhondzadeh2013), anti-epileptic drugs (Nakamura et al., Reference Nakamura, Mihara, Nagai, Suzuki and Kondo2009), and non-pharmacological treatment. The most common combination is the combination of antipsychotics and antidepressants (Mao & Zhang, Reference Mao and Zhang2015). A meta-analysis found that, in Western countries, approximately 30% (range: 11–40%) of schizophrenia patients take antidepressants (Mao & Zhang, Reference Mao and Zhang2015). In clinical care, antidepressants are increasingly used in patients with schizophrenia, most commonly to target negative, cognitive, and depressive symptoms. The potential therapeutic role of antidepressants in schizophrenia is supported by studies (Cornblatt et al., Reference Cornblatt, Lencz, Smith, Olsen, Auther, Nakayama and Correll2007; Fusar-Poli et al., Reference Fusar-Poli, Frascarelli, Valmaggia, Byrne, Stahl, Rocchetti and McGuire2015) that suggest that antidepressants can reduce the transition from a high risk of clinical psychosis to presenting psychosis, and that they can reduce psychosis relapse in patients with schizophrenia and postpsychotic depression (Siris et al., Reference Siris, Bermanzohn, Mason and Shuwall1994).
Antidepressants are primarily classified into categories such as monoamine oxidase inhibitors, 5-hydroxytryptamine (5-HT) and norepinephrine (NE) reuptake blockers, selective serotonin reuptake inhibitors, serotonin antagonists and reuptake inhibitors, serotonin modulators and agonists, and NE and dopamine reuptake inhibitors. The diverse mechanisms of these drugs suggest their potential in treating schizophrenia’s negative symptoms. Past research has suggested that antidepressants may influence schizophrenia’s negative symptoms through their effects on γ-aminobutyric acid (GABA) and glutamate neurotransmitters (Pehrson et al., Reference Pehrson, Jeyarajah and Sanchez2016), as well as their action on multiple neurotransmitters like dopamine. A key characteristic of schizophrenia is the altered activity of the dopamine system in different brain regions, which gives rise to various symptom profiles. Antidepressants also target post-synaptic 5-hydroxytryptamine type 3 (5-HT3) receptors and presynaptic α2-adrenergic receptors and indirectly stimulate post-synaptic 5-hydroxytryptamine (serotonin) receptor 1A (5-HT1A) receptors (de Boer, Reference de Boer1996), all of which are implicated in the pathophysiology of schizophrenia. Preceding evidence suggests that 5-HT3 receptor antagonists hold promise for treating schizophrenia due to their potential therapeutic benefits (Costall & Naylor, Reference Costall and Naylor1992).
Previous meta-analyses have indicated that the combination of antipsychotic and antidepressant treatment may be effective for the negative symptoms of schizophrenia (Galling et al., Reference Galling, Vernon, Pagsberg, Wadhwa, Grudnikoff, Seidman and Correll2018; Rummel et al., Reference Rummel, Kissling and Leucht2006). The driving factor for overall symptom improvement is the negative symptoms, rather than the reduction of positive symptoms or total symptom severity. It was also found that antidepressants have demonstrated an advantage in improving negative symptoms in studies involving the addition of first-generation antipsychotics, but this was not confirmed in studies involving the addition of second-generation antipsychotics.
However, so far, the efficacy of antidepressants in the treatment of negative symptoms of schizophrenia is still controversial. Although most studies have shown that antidepressants are effective in the treatment of negative symptoms of schizophrenia, several studies have shown that antidepressants cannot improve negative symptoms of schizophrenia (Sepehry et al., Reference Sepehry, Potvin, Elie and Stip2007; Singh et al., Reference Singh, Singh, Kar and Chan2010; Usall et al., Reference Usall, López-Carrilero, Iniesta, Roca, Caballero, Rodriguez-Jimenez, Oliveira, Bernardo, Corripio, Sindreu, Piqueras, Felipe, de Corres, Ibáñez and Huerta2014). These contradictory results may be accounted for by methodological limitations: small sample, short trial length, a range of concomitant antipsychotic regimens, or lack of exclusion of subjects with depressive symptoms. Given the current scenario, the effectiveness of antipsychotic medications in addressing the negative symptoms of schizophrenia is restricted, prompting a shift toward combined therapies, where the use of adjunctive antidepressants emerges as a beneficial strategy.
In this meta-analysis, we conducted a comprehensive review of recent studies on the adjunctive use of antidepressants in treating negative symptoms of schizophrenia. Previous meta-analyses did not adequately focus on negative symptoms in the inclusion criteria, which may lead to bias in the research results. However, in this study, we specifically focused on negative symptoms and made them a key aspect of the inclusion criteria. The improvement in the inclusion criteria lies in our clear emphasis on the assessment of schizophrenia’s negative symptoms and prioritizing their evaluation. We also ensured that the scales used to assess negative symptoms are flexible and can accommodate different measurement methods. By incorporating this focus, we aim to provide more detailed and targeted insights into the effects of interventions on negative symptoms, which are often underrepresented in schizophrenia research. At the same time, because there are common features between negative symptoms and depressive symptoms, it is necessary to focus on depressive symptoms, as detailed in Supplementary Table S1. Consequently, we systematically reviewed the effects of antidepressant use in patients with negative symptoms and incorporated it as an essential criterion for selection. Our objective is to determine which antidepressant exhibits superior efficacy compared to placebo while also exploring potential differences among various antidepressants and to provide valuable insights for future therapeutic approaches targeting negative symptoms in individuals with schizophrenia.
Methods
Search and selection criteria
In this systematic review and meta-analysis, we thoroughly examined the reference lists of previous systematic reviews and meta-analyses pertaining to this topic within our research group, while also conducting an extensive search. Our literature search encompassed Web of Science (from its inception until April 16, 2025) and PubMed (from its establishment until April 16, 2025). We conducted a comprehensive search employing various combinations of keywords pertaining to the three central themes: negative symptoms, schizophrenia, and antidepressants. The specific search terms and search strategy can be found in the Supplementary Materials.
Inclusion criteria for the study: (1) patients aged 18 to 65 years; (2) clearly diagnosed with schizophrenia, regardless of gender, race, or diagnostic criteria used; (3) the inclusion criteria or primary outcome measure specifically focus on the negative symptoms of schizophrenia, without restriction on the scales used to assess negative symptoms; (4) the inclusion criteria focus on depressive symptoms (regardless of the scale used) and exclusions were made; (5) randomized controlled trials, including studies on all existing antidepressants and placebo treatments on the market; (6) written in English. exclusion criteria: (1) studies involving comorbid psychiatric disorders and substance abuse; (2) abstracts, reviews, case reports, commentaries, letters, and protocols; (3) studies from which valid data cannot be obtained after contacting the corresponding author and foreign experts for literature searches. The primary outcome measure focused on assessing changes in negative symptoms of schizophrenia using either the scale for Assessment of Negative Symptoms (SANS; Andreasen, Reference Andreasen1982) or the scale for Positive and Negative Symptoms-Negative Subscale (PANSS-N; Kay et al., Reference Kay, Fiszbein and Opler1987, Reference Kay, Opler and Lindenmayer1988). These scales have demonstrated high internal consistency and external validity among various population groups; moreover, they are considered relatively comparable (Czobor et al., Reference Czobor, Bitter and Volavka1991; McAdams et al., Reference McAdams, Harris, Bailey, Fell and Jeste1996; Thiemann et al., Reference Thiemann, Csernansky and Berger1987).
Outcome measures and data extraction
The primary outcome was the mean change from baseline to endpoint in negative symptoms of schizophrenia as measured by negative subscales of the PANSS-N (Fusar-Poli et al., Reference Fusar-Poli, Papanastasiou, Stahl, Rocchetti, Carpenter, Shergill and McGuire2015; Kay et al., Reference Kay, Fiszbein and Opler1987) or the SANS (Andreasen, Reference Andreasen1989). If the change data were not available, we used the mean score at the study endpoint of these scales. The articles’ quality was classified according to the risk of bias tool (Cumpston et al., Reference Cumpston, Li, Page, Chandler, Welch, Higgins and Thomas2019). The selected literature was assessed based on seven indicators: random sequence generation, allocation concealment, blinding of investigators and subjects, blinded evaluation of study outcomes, completeness of outcome data, selective reporting of study outcomes, and other sources of bias. If all criteria were met with low risk factors present or accounted for appropriately in case there were no high-risk indicators identified, then it was determined that the study had a low risk of bias. In case one or more criteria remained unclear without any high-risk indicators being found, then it was considered that the study had an unclear risk of bias. However, if one or more indicators were deemed as high risk, then it was concluded that the study had a high risk of bias. Two resident psychiatrists from the department of psychiatry independently reviewed the results of the updated search, obtained full-text articles, assessed eligibility criteria, and extracted data. The data analysis was performed by one of the psychiatrists, who then verified and checked the results with another psychiatrist. If an article meets the inclusion criteria but complete data cannot be extracted through various methods, such as contacting the corresponding author or inviting foreign experts to conduct literature searches, the article will be excluded in order to minimize the result bias caused by missing data. Any discrepancies were resolved through discussion.
Statistical analysis
This network meta-analysis (NMA) focuses on the measurement methods for the reduction of negative symptom scores. Considering the different scales used in the included studies to measure the reduction of negative symptoms, the overall effect size for the primary outcome measure was calculated using the standardized mean difference (SMD). The meta-analysis was conducted by combining the standardized effect sizes between trials using a random-effects model, which accounted for the heterogeneity between studies due to variability in case combinations and trial settings. We performed a sensitivity analysis, and the results showed that no study had an exceptionally large impact on the pooled effect size, indicating that our effect size remained stable. Global and local inconsistency analyses were performed using Stata MP 17 (64-bit edition) to assess the viability of integrating direct and indirect comparisons across different scales. Heterogeneity was assessed utilizing the chi-square test, I 2 statistic, and forest plots, with I2 values ranging from 0% to 100% (where smaller values indicate lower heterogeneity). A p-value below 0.05 denoted a significant reduction in heterogeneity. The NMA encompassed the creation of network diagrams, pairwise comparison forest plots, funnel plots, and cumulative ranking curves. The network diagram depicts the interrelationships among various intervention measures, where each node symbolizes a distinct intervention, and each edge signifies a direct comparison between two interventions. Moreover, the size of a node reflects the frequency of the corresponding drug as a comparator in the study. The thickness of edges connecting two nodes is directly related to the number of studies that include the two directly compared intervention measures. The funnel plot and Egger’s test were used to assess publication bias. Surface under the cumulative ranking curve (SUCRA) and average ranking were employed to evaluate each outcome. The greater the area under the curve for each drug, the higher the likelihood that it signifies the optimal intervention.
Results
Search results
The literature search initially yielded 1890 potentially relevant studies. After removing duplicates, 1483 studies remained. Of these 1483 studies, 1436 were excluded after title and abstract screening, leaving 47 studies for further full-text review (reasons for exclusion are shown in Figure 1). Finally, 15 articles (n = 655) met the inclusion criteria and were included in this systematic review and meta-analysis.
Flow chart of the systematic literature search and excluded and included studies.
Characteristics of the included studies
Table 1 lists the detailed information of all included studies. Among 481 patients with gender specified, 312 were male (65%), and the median trial duration was 8 weeks (range: 4–48 weeks). The risk of bias assessment is shown in Supplementary Figure S1. Several trials lacked clear details on the randomization procedure and allocation concealment. In about 40% of the studies, the blinding of patients and staff remained unclear. However, there was no indication of high risk related to insufficient blinding. The risk of bias in outcome assessment related to blinding showed similar uncertainty (40%), but did not suggest any high risk.
Included studies, patients, and treatment characteristics
Note: AD, antidepressant; AP, antipsychotic; bl, baseline; BPRS, brief psychiatric rating scale; BPRS-P, BPRS positive subscale; BUP, Bupropion; CIT, citalopram; CLO, clozapine; d, days; DB/RCT, Double-blind, randomized, placebo-controlled trial; DUI, duloxetine; ESC, escitalopram; FGA, first-generation antipsychotic; FLU, fluoxetine; FLV, fluvoxamine; HAL, haloperidol; HDRS, Hamilton depression rating scale; MDD, major depressive disorder; MIR, mirtazapine; NR, not reported; OLZ, olanzapine; PANSS, positive and negative syndrome scale; PANSS-N, PANSS negative subscale; PAR, paroxetine; REB, reboxetine; RIS, risperidone; SANS, scale for the assessment of negative symptoms.
Figure 2 shows the qualified comparison network for the main outcomes. The identified studies included drugs vortioxetine, fluvoxamine, paroxetine, mirtazapine, escitalopram, reboxetine, citalopram, fluoxetine, duloxetine, bupropion, placebo. Fifteen of the included studies met our criteria for patients with negative symptoms and included treatments: vortioxetine (N = 1), fluvoxamine (N = 2), paroxetine (N = 1), mirtazapine (N = 2), escitalopram (N = 1), reboxetine (N = 3), citalopram (N = 3), fluoxetine (N = 2), duloxetine (N = 1), bupropion (N = 1). Figure 2 shows the relationships between 10 different interventions, with each blue dot representing an intervention and each black line representing a direct comparison between two interventions. Based on Figure 2, the blue dot for the reboxetine group has the largest area. Therefore, in the included studies, the reboxetine group was the most commonly compared group. Based on the width of the black lines connecting the two blue dots, which represent the proportion of studies comparing the two interventions directly, the black line between reboxetine and placebo is the thickest, indicating that the number of studies comparing reboxetine and placebo directly was the highest. Similarly, this also occurs in the direct comparison of citalopram and placebo.
Network geometry. Among them, PLA represents placebo, VOR represents vortioxetine, FLV represents fluvoxamine, PAR represents paroxetine, MIR represents mirtazapine, EsCIT represents escitalopram, REB represents reboxetine, CIT represents citalopram, FLU represents fluoxetine, DUL represents duloxetine, and BUP represents bupropion. The figure shows the results of a network meta-analysis that directly compares the 10 interventions. The width of the lines is proportional to the number of lines comparing each pair of interventions, and the size of each blue dot is proportional to the sample size of the interventions.
Efficacy of treatments for negative symptoms
Ten drugs were compared with a placebo (Table 2). For mirtazapine, two placebo-controlled trials were conducted. Mirtazapine significantly outperformed placebo in treating negative symptoms (N = 2, n = 48, SMD −1.73, CI −2.60, −0.87). Based on one trial, duloxetine significantly outperformed placebo in negative symptoms (N = 1, n = 64, SMD −1.19, CI −2.17, −0.21). In direct comparisons between antidepressants, significant differences were found between mirtazapine and reboxetine, citalopram, and bupropion, while no significant differences were observed in direct comparisons between other antidepressants or between antidepressants and placebo. We also made predictions regarding the efficacy of antidepressants, with the area under the curve representing the efficacy ranking of antidepressants in treating negative symptoms, as shown in Supplementary Figure S2. It can be seen that mirtazapine (MIR) has the largest area under the curve, followed by duloxetine (DUL). The area under the curve for bupropion (BUP) is nearly identical to that of placebo (PLA). This indicates that mirtazapine and duloxetine are more effective in treating negative symptoms of schizophrenia compared to other antidepressants, while bupropion is less effective than other antidepressants. Table 3 shows the cumulative probability prediction values for the ranking of each drug, with a higher probability indicating better drug efficacy. The probability of mirtazapine being effective is the highest at 93.9%, followed by duloxetine at 76.2%.
Results of network meta-analysis of the improvement rate of negative symptoms of schizophrenia (SMD value and 95% CI)
Note: BUP, bupropion; CIT, citalopram; DUL, duloxetine; EsCIT, escitalopram; FLU, fluoxetine; FLV, fluvoxamine; MIR, mirtazapine; PAR, paroxetine; PLA, placebo; REB, reboxetine: SMD, the standardized mean difference; and VOR, vortioxetine.
Cumulative probability predicted values for each drug
Note: The SUCRA represents the area under the curve, indicating better drug efficacy with a larger area. Best probability ranks drug efficacy as a percentage, with higher percentages indicating better efficacy. When there is a discrepancy, the area under the curve result should be used. BUP, bupropion; CIT, citalopram; DUL, duloxetine; EsCIT, escitalopram; FLU, fluoxetine; FLV, fluvoxamine; MIR, mirtazapine; PAR, paroxetine; PLA, placebo; REB, reboxetine; and VOR, vortioxetine.
Quality evaluation
Supplementary Figure S1 summarizes the results of the quality assessment. All 15 studies were randomized controlled clinical trials (RCTs). The included RCTs were assessed using the Cochrane Collaboration’s tool for assessing the risk of bias. Ten studies had a low risk of bias, and 5 studies had an unclear risk of bias. The quality assessment results of each study were statistically different. All the 15 studies were blinded, of which 12 were double-blind and 3 were triple-blind.
Publication bias
The funnel plot (Supplementary Figure S3) illustrates that studies with smaller sample sizes are symmetrically arranged at the base of the graph, while those with larger sample sizes are distributed toward the top of the funnel plot and converge toward the center of the graph. The overall symmetrical distribution of the plot suggests that the results of the included studies are relatively consistent, with no significant asymmetry observed, indicating a low risk of publication bias. The funnel plot provides a preliminary assessment of publication bias; however, to further evaluate this, we conducted Egger’s test. The results of Egger’s test, with a p-value of 0.342 (p > 0.05), suggest a low risk of publication bias. Combining the results from the funnel plot and Egger’s test, it can be concluded that the risk of publication bias in the included studies is low.
Inconsistency test
Stata software was used to test the inconsistency of statistical results. From the pairwise comparison forest plot in Supplementary Figure S4, we can see that the p value is 0.258, which is >0.05, indicating that the heterogeneity test is not significant. A closed-loop structure was formed in the network diagram by direct and indirect comparisons, so we conducted a circular inconsistency test for this study (Chaimani & Salanti, Reference Chaimani and Salanti2015), as shown in Supplementary Figure S5. According to Supplementary Figure S5, the results of the node segmentation method show that there is a direct comparison between reboxetine and citalopram, which forms a closed loop. The 95% confidence interval for the closed loop is (0.00, 1.55) and the IF is 0.76. There was no significant difference between the direct and indirect results. Thus, the inconsistency test in our NMA is not significant, which indicates that our results analysis is reliable.
Discussion
To the best of our knowledge, this study represented the first NMA examining the efficacy of adjunctive antidepressant medication in treating negative symptoms. The results of this meta-analysis indicated that antidepressant medication is superior to placebo in improving overall symptoms, driven by negative symptoms rather than reducing positive or general symptoms. We concentrated our efforts on randomized placebo-controlled trials, encompassing a total of 344 patients in the treatment group and 311 patients in the placebo group. While previous systematic reviews and meta-analyses have touched upon the supplementary treatment of negative symptoms with antidepressants (Fusar-Poli et al., Reference Fusar-Poli, Papanastasiou, Stahl, Rocchetti, Carpenter, Shergill and McGuire2015; Sepehry et al., Reference Sepehry, Potvin, Elie and Stip2007), they lacked specific and stringent inclusion criteria specific to negative symptoms. We also encourage future research to pay more attention to patients with significant negative symptoms or primary negative symptoms and mild positive symptoms.
In the present meta-analysis, we observed that mirtazapine and duloxetine may be superior to other antidepressants in treating negative symptoms. This may be due to some similarities in their mechanisms of action. Mirtazapine has interesting psychopharmacological characteristics: It antagonizes the 5-HT2 and 5-HT3 receptors, indirectly activates the 5-HT1A receptor, and antagonizes the adrenergic a2 receptor (Croom et al., Reference Croom, Perry and Plosker2009; Schreiber et al., Reference Schreiber, Bleich and Pick2002). Duloxetine selectively inhibits the reuptake of serotonin and NE. Serotonin is a key neurotransmitter in schizophrenia: It is associated with the negative and cognitive symptoms of schizophrenia (Fukuda, Reference Fukuda2014). Reduced 5-HT1A receptor binding in the amygdala is related to specific components of the negative symptoms of schizophrenia (Yasuno et al., Reference Yasuno, Suhara, Ichimiya, Takano, Ando and Okubo2004). Literature reviews also show the role of NE in the pathogenesis of negative symptoms in schizophrenia (Yamamoto & Hornykiewicz, Reference Yamamoto and Hornykiewicz2004). The effects of mirtazapine and duloxetine on serotonin and NE validate the potential mechanisms for treating the negative symptoms of schizophrenia. Consequently, for the future management of negative symptoms in schizophrenia, it is advisable to commence treatment with antipsychotic medications alone. If the initial response is unsatisfactory, considering potential drug interactions, the addition of mirtazapine or duloxetine may be considered as adjunctive treatment for negative symptoms. In the included RCTs, we observed that the minimum effective dose of mirtazapine was 30 mg/d and for duloxetine, 60 mg/d. At the same time, the gradual improvement of negative symptoms during treatment can be monitored to better assess the efficacy of antidepressant therapy.
In 2010, Singh et al. (Reference Singh, Singh, Kar and Chan2010) conducted a meta-analysis aimed at assessing the efficacy of adjunctive antidepressant medication in treating negative symptoms among patients with schizophrenia. The analysis encompassed 23 trials, involving a total of 819 patients. The findings of this meta-analysis revealed the effectiveness of antidepressant medication in alleviating negative symptoms in schizophrenia. Specifically, fluoxetine, trazodone, and ritanserin demonstrated beneficial effects, whereas mirtazapine, reboxetine, mianserin, fluvoxamine, sertraline, paroxetine, and citalopram did not exhibit any favorable effects. There is a difference in conclusions between our study and the aforementioned meta-analysis, which may be due to the fact that this meta-analysis particularly emphasized the importance of negative symptoms in its inclusion criteria and increased the focus on depressive symptoms. Compared to previous studies, we conducted a more rigorous and detailed screening of the study population and sample. This may have led to differences in sample characteristics and baseline data, which could affect the results. Furthermore, this meta-analysis included the latest studies published in recent years, whereas previous studies may not have covered these newer pieces of evidence. As research evidence accumulates and is updated, the results may change. Additionally, publication bias could also affect the consistency of results across different studies. However, overall, the use of antidepressants is considered effective in treating negative symptoms.
This meta-analysis aligns with the recent study conducted by (Galling et al., Reference Galling, Vernon, Pagsberg, Wadhwa, Grudnikoff, Seidman and Correll2018), which investigated the adjunctive treatment of schizophrenia using antipsychotic and antidepressant medications. Galling et al. found that mirtazapine demonstrated relatively superior efficacy in addressing negative symptoms when compared to other antidepressant medications, a finding that is in line with the results of this study. It should be noted that this study specifically concentrates on evaluating the efficacy of interventions for negative symptoms (Jockers-Scherübl et al., Reference Jockers-Scherübl, Bauer, Godemann, Reischies, Selig and Schlattmann2005; Moazen-Zadeh et al., Reference Moazen-Zadeh, Bayanati, Ziafat, Rezaei, Mesgarpour and Akhondzadeh2020). While our findings align with previous research conducted by Galling et al., neither our study nor theirs was able to distinguish between primary and secondary negative symptoms (e.g. those stemming from anxiety, depression, paranoia, extrapyramidal symptoms, or sedation), nor could we differentiate persistent negative symptoms (‘deficit symptoms’) from transient ones (Carbon & Correll, Reference Carbon and Correll2014; Carpenter et al., Reference Carpenter, Heinrichs and Wagman1988). However, this does not necessarily imply the clinical significance of these effects. First, it is vital to consider that the observed improvement in negative symptoms in most trials might partially stem from the amelioration of secondary negative symptoms. As a result, these findings may potentially overestimate the actual therapeutic efficacy. Second, even though a slight change in symptom assessment scales may be observed, it may not directly translate into an impact on patients’ functional capacity or quality of life. However, the therapeutic effectiveness could also be influenced by the duration of the study and the stage of patients’ disease progression. In this meta-analysis, the included studies predominantly involved patients with a course of illness concentrated around 10–20 years, indicating generally longer durations of illness. Thus, it was difficult for us to determine the potential changes in the efficacy of antidepressants in the early stages of schizophrenia in this meta-analysis. Since negative symptoms are a chronic aspect of schizophrenia, clinical trials must be sufficiently long to demonstrate that the observed treatment response is not temporary. Moreover, certain specific negative symptoms that need to be addressed depend on interactions with the environment, and it may take months to observe measurable and stable improvements. There is a growing consensus that trials should last approximately 6 months to assess the efficacy of treatments for negative symptoms (Laughren & Levin, Reference Laughren and Levin2006). However, to our knowledge, there are few published trials with a follow-up period of 6 months. This may be due to the concern that if patients are randomly assigned to receive placebo adjunctive treatment for such an extended period, their negative symptoms, as well as depressive symptoms that may be confused with negative symptoms, could worsen. This creates significant ethical and feasibility obstacles in the design of trials that include a placebo treatment group. Furthermore, the interpretation of clinical outcomes may be confounded. Loss of clinical stability may lead to an increase in paranoid or other symptoms of schizophrenia, which are behaviorally similar to negative symptoms, such as social withdrawal. Conversely, the emergence of psychotic agitation may be confused with a reduction in negative symptoms (Alphs, Reference Alphs2006). In this meta-analysis, we found that duloxetine and mirtazapine showed relatively better efficacy in the adjunctive treatment of schizophrenia negative symptoms for 8 weeks compared to other antidepressants. However, we cannot assess whether the results of long-term adjunctive treatment for negative symptoms would be stable and measurable. The variability in treatment duration among the trials included in this study limits the generalizability of the findings. Therefore, future research should focus on long-term adjunctive antidepressant treatment for schizophrenia negative symptoms to better confirm the efficacy of antidepressants in treating negative symptoms in schizophrenia.
Currently, distinguishing between primary and secondary negative symptoms remains a complex task, underscoring the need for future studies to delve into this topic with finer granularity and a more comprehensive approach. This meta-analysis further validates the effectiveness of antidepressant medications in the adjunctive treatment of negative symptoms, providing valuable insights for future research in this field. The positive and negative symptoms, as well as cognitive impairments, are core features of schizophrenia. Negative symptoms may manifest as a lack of motivation and the absence of elements that constitute normal social and emotional responses, but these are also common in depression (Malla, Reference Malla1995). Siris et al., Reference Siris, Adan, Cohen, Mandeli, Aronson and Casey1988 found that half of the patients with post-psychotic depression also met the criteria for negative symptoms, suggesting a syndromic overlap between depression and negative symptoms. Anhedonia, emotional blunting, irritability, motivation, sociability, and loneliness appeared in both the negative and depressive domains (Romney & Candido, Reference Romney and Candido2001; Sax et al., Reference Sax, Strakowski, Keck, Upadhyaya, West and McElroy1996; Whiteford et al., Reference Whiteford, Riney and Csernansky1987). A review (Krynicki et al., Reference Krynicki, Upthegrove, Deakin and Barnes2018) indicated that, despite considerable heterogeneity in research findings, certain depressive features (such as low mood, suicidal thoughts, and pessimism) differ from negative symptoms of schizophrenia, which are characterized by aphasia and emotional flatness. For instance, (Liemburg et al., Reference Liemburg, Castelein, Stewart, van der Gaag, Aleman and Knegtering2013; Messinger et al., Reference Messinger, Trémeau, Antonius, Mendelsohn, Prudent, Stanford and Malaspina2011) found that while depressive mood and suicidal intent reflect the depressive domain, emotional blunting and aphasia belong to the ‘expressive deficit’ domain of negative symptoms. The contribution of depression to negative symptoms is a possibility, and improvement of depression through antidepressant medications can influence the change in negative symptoms. It is a continuing debate whether the improvement is secondary to the reduction of depressive symptoms (Rummel et al., Reference Rummel, Kissling and Leucht2005). All studies included in this meta-analysis used standardized tools to assess negative symptoms, and changes in scores after intervention were standardized. This is primarily aimed to address the core issues of negative symptoms. Although this meta-analysis screened for depressive symptoms and focused on negative symptoms, the final observed effect may reflect improvements in depressive symptoms that were mistakenly attributed to negative symptoms. This is due to factors such as the inconsistency of assessment scales for depressive and negative symptoms across studies, as well as the variability in evaluation standards among experimenters.
When evaluating efficacy, it is important to not rely only on changes in scale scores. Given the high prevalence and significant burden of schizophrenia, particularly with regard to negative and depressive symptoms (Bobes et al., Reference Bobes, Arango, Garcia-Garcia and Rejas2010; Buckley et al., Reference Buckley, Miller, Lehrer and Castle2009; Conley et al., Reference Conley, Ascher-Svanum, Zhu, Faries and Kinon2007; Fervaha et al., Reference Fervaha, Foussias, Agid and Remington2014; Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005; Mäkinen et al., Reference Mäkinen, Miettunen, Isohanni and Koponen2008; Olivares et al., Reference Olivares, Sermon, Hemels and Schreiner2013; Rosen et al., Reference Rosen, Miller, D’Andrea, McGlashan and Woods2006; Sands & Harrow, Reference Sands and Harrow1999; Tapp et al., Reference Tapp, Kilzieh, Wood, Raskind and Tandon2001), even if a small proportion of patients exhibit symptom improvement, it can still be deemed clinically meaningful, as we have estimated. Moreover, beyond the effectiveness of interventions and the quality of supporting evidence, the perspectives of patients should also be taken into account (Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005).
Limitations
In the studies included in this meta-analysis, we focused more on research that screened for depressive symptoms in the inclusion criteria, regardless of the scale used to assess depressive symptoms. This refinement led to resulting in a relatively low proportion of depression in the overall psychopathology. While acknowledging the novelty and significance of our study, it is imperative to take into account the following limitations. (i) The number of RCTs comparing antidepressants with placebo in augmenting baseline antipsychotic treatment is relatively small, which precludes a more detailed analysis of individual antipsychotic-antidepressant combination therapies. (ii) Many of the included studies are small scale, so our results may be influenced by small study effects, including exaggerated effect sizes (Cumpston et al., Reference Cumpston, Li, Page, Chandler, Welch, Higgins and Thomas2019). However, since the risk of bias assessment is acceptable, this overestimation is likely to be relatively small. Clearly, large randomized controlled trials with sufficient power would help provide more definitive evidence. (iii) The types of antidepressants included in this study were based on existing randomized controlled trials, and the variety of antidepressants is not comprehensive. Therefore, future research on the treatment of negative symptoms of schizophrenia should focus on currently confirmed antidepressant medications that are effective for negative symptoms and collect more reliable and extensive data for further investigation. (iv) The included studies used various scales to assess negative symptoms. Although we made efforts to standardize the results, potential errors were not fully eliminated. (v) The included studies could not clearly specify the types of antipsychotic medications used in the background treatment, making it impossible to conduct further analysis.
Conclusions
To the best of our knowledge, this study represents the first investigation into the use of antidepressants for treating negative symptoms in schizophrenia. In the current meta-analysis, we observed that mirtazapine and duloxetine exhibited the highest efficacy in the treatment of negative symptoms in patients with schizophrenia among antidepressants. The negative symptoms in schizophrenia patients present a significant challenge in current treatment approaches for the disorder. Despite the numerous proposed treatments for negative symptoms, their efficacy remains uncertain. Among these, antidepressants have been the most commonly suggested; however, their clinical application remains relatively limited. This research introduces a novel perspective on utilizing antidepressants for addressing negative symptoms in schizophrenia, potentially aiding in the identification of optimal treatment strategies. In this meta-analysis, the articles we included used different scales for the primary outcome measure, so although we used standardized mean differences (SMD) to eliminate their impact on the results, there still exist differences in the comprehensiveness of the negative symptom assessment between the two scales. Compared to PANSS-N, SANS represents more items; thus, SANS may provide more interpretable factors for discovery (Blanchard & Cohen, Reference Blanchard and Cohen2006). In future research, we propose the adoption of consistent scales for the evaluation of related indicators.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291725101803.
Acknowledgements
This study was supported in part by Sichuan Provincial Department of Science and Technology (2023NSFSC0124, 2022YFS0615), Luzhou Science and Technology Bureau-Southwest Medical University (2021LZXNYD-D04), Luzhou Science and Technology Bureau (2022-SYF-96, 2022-ZRK-186), Zigong Science and Technology Bureau (2021YXY03, 2022ZCNKY02), High quality development project of Zigong City Hospital (ZG-KY-2023-047, ZG-PT-2023-026), Youth Project of Southwest Medical University (2021ZKQN064), Youth Project of Affiliated Hospital of Southwest Medical University (2017-PT-9, 2011 [37] and 16009), and Central Nervous System Drug Key Laboratory of Sichuan Province (200029-01SZ).
Competing interests
The authors declare none.
Open access
