Peripheral serotonin levels as a predictor of antidepressant treatment response: A systematic review

that higher baseline 5-HT was associated with a greater antidepressant response to SSRIs, prompting future studies to investigate this hypothesis. Our results also confirm previous reports that SSRI treatment is associated with a decrease in peripheral 5-HT levels; however, we were not able to confirm that larger decreases of 5-HT are associated with better treatment outcome as results were inconclusive.


Rationale
Despite the global estimate of over 300 million individuals affected by depression and its status as the leading cause of non-fatal health loss (WHO, 2017), there remains considerable potential for enhancing the treatment for depressive disorders.Currently, only about 40-60% of patients respond to first-line antidepressant treatments (Perna et al., 2020), and fewer than one third of patients with major depressive disorder (MDD) achieve remission after the first treatment attempt (Trivedi et al., 2006).Efforts to enhance the effectiveness of antidepressant treatment have been ongoing for decades.These efforts have been hampered by the ambiguous mechanisms underlying the effects of antidepressant medications.
Currently, there are no reliable predictors of antidepressant treatment response.The selection of medication is primarily guided by clinical guidelines and clinicians' preferences, incorporating an element of trial and error.Biomarkers capable of predicting treatment response could facilitate personalized prescribing of antidepressant medication, saving time and resources while significantly improving symptom control and minimizing side effects and suffering.
The earliest developed antidepressant medications targeted monoamines, such as 5-hydroxytryptamine (5-HT; serotonin), norepinephrine (NE) or dopamine (DA), presumably by blocking neurotransmitter reuptake via the transporter proteins or by blocking the degradation of neurotransmitters in the synapse (Hamon and Blier, 2013).Among the monoamines, 5-HT has been extensively investigated in the context of depression (Morrissette and Stahl, 2014).5-HT is synthesized from tryptophan in the enterochromaffin cells in the gut and in neurons in the central nervous system, as 5-HT cannot cross the blood-brain barrier (Szeitz and Bandiera, 2018).The largest quantities by far of circulating peripheral 5-HT can be found within the dense granules of platelets, into which they are taken up after being released from the enterochromaffin cells into the blood stream (Szeitz and Bandiera, 2018).Platelet activation, triggered by signals such as damaged endothelium and ischemia, leads to the release of 5-HT from platelets into the plasma (Brand and Anderson, 2011;Szeitz and Bandiera, 2018).
While the monoamine theory of depression does not sufficiently explain the entire pathophysiology of MDD (Moncrieff et al., 2022), monoamines remain relevant in understanding and predicting treatment response.Previous studies suggest that a certain degree of 5-HT system activity may be required for optimal antidepressant efficacy of medications such as SSRIs (selective serotonin reuptake inhibitors); for instance, patients with remitted MDD who undergo tryptophan depletion are more prone to relapse if they are taking serotonergic antidepressants compared to those without antidepressant treatment or those on antidepressants with different mechanisms of action (Jesulola et al., 2018;Ruhe et al., 2007).Some studies have reported that a larger decrease of 5-HT during SSRI treatment was associated with better treatment outcome (Gupta et al., 2016;Holck et al., 2019), although it is unclear if this finding is robust across studies.
Previous research suggests that tryptophan metabolites, including 5-HT, may be potential biomarkers of depression and specifically to predict antidepressant treatment response (Guerreiro Costa et al., 2022).Although the association between peripheral blood 5-HT and central levels of 5-HT is uncertain (Audhya et al., 2012;Gao et al., 2008), previous studies have demonstrated a link between peripheral 5-HT and antidepressant activity.Identifying associations between circulating 5-HT levels and antidepressant response could enhance treatment prediction and expand our understanding of the biological mechanisms behind antidepressant treatment.To our knowledge, no study has yet systematically summarized the evidence for baseline peripheral 5-HT as a predictor of antidepressant treatment response.

Objectives
To systematically review the evidence for an association between baseline peripheral blood 5-HT levels and antidepressant treatment response in patients with depressive disorders.As a secondary objective, we also explore associations between changes in peripheral 5-HT levels during antidepressant treatment and treatment outcome in this population.

Eligibility criteria
We conducted a systematic search of the literature to identify all studies that measured 5-HT levels in peripheral blood (whole blood, serum, plasma or platelets) before the initiation of a pharmacological treatment for depressive disorders as defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM) (Diagnostic and Statistical Manual of Mental Disorders, 2013) or the International Classification of Diseases (ICD) ("International statistical classification of diseases and related health problems (10th ed.),", 2019) and that also examined the relationship between baseline levels of 5-HT and the outcomes of the antidepressant treatment.Inclusion criteria required the studies to be full reports of original research, in English and involving human subjects.Studies that included subjects of all different ages were considered for inclusion.Studies were included if the subjects had comorbid illnesses; however, the depressive disorder should be specified as the primary diagnosis.A pharmacological intervention was considered eligible if the study classified it as possessing antidepressant properties, even when added to other medications.We excluded studies that only investigated the effect of psychotherapy, electroconvulsive therapy (ECT) or other non-pharmacological treatments.We included studies regardless of length of treatment, wash-out and follow-up.We included both randomized and non-randomized intervention studies.
The protocol was registered at PROSPERO (8 April 2023, number CRD42023411877).The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Guidelines were followed (Page et al., 2021).
Following our preliminary searches, we broadened our inclusion criteria to encompass populations defined by the Research Diagnostic Criteria (RDC) (Spitzer et al., 1978) in addition to DSM and ICD criteria.The protocol was updated accordingly.

Information sources and search strategy
We conducted our literature search in the databases MEDLINE, Embase, PsycINFO, and the Cochrane Central Register of Controlled Trials.Search terms included Boolean combinations of the following: depressive disorders, major depressive disorder; SSRIs, SNRIs, antidepressive agents, antidepressants; serotonin, 5-hydroxytryptamine; antidepressant response, treatment response, treatment outcome, remission.The specific search strategies were created by a health sciences librarian specializing in systematic review searching together with the review team (AH and DL).The initial search was performed on January 2023 including all relevant papers published up to that date.The search was re-run on 20 November 2023 prior to final analysis.

Selection process
Two authors (AH and one of MA, PM, ÅW and DL) independently conducted initial screening of records based on title and abstract.Articles falling outside the scope of our review, and those not written in English, were excluded.Subsequently full-text records were evaluated for inclusion independently by the same pairs of authors.Disagreements were resolved through discussion until consensus was reached.The screening process is illustrated in Fig. 1.

Data collection process and data items
Two reviewers (AH and one of MA, PM, ÅW and DL) extracted data from each selected report independently.Collected information included information on study aim, design, baseline population characteristics, inclusion and exclusion criteria, withdrawals and exclusions, diagnostic criteria, interventions, outcomes of depression and 5-HT levels, key conclusions of study authors, study funding and conflicts of interest.Data from all reported time points were collected.Authors were contacted for additional information when necessary.

Assessment of quality
The quality of the included research reports was assessed using a modified version of seven criteria proposed by Strawbridge et al., (Strawbridge et al., 2015), which, in turn, had been adapted from the criteria developed by the Evidence-Based Medicine Working Group that had been modified for use in prognostic investigations (Fekadu et al., 2009).With these criteria, studies can score positively (+1), negatively (− 1) or with no change of score on each criterion resulting in a quality score ranging from − 7 to 7. The following domains were assessed: cohort formation, sample size, trial/follow-up length, collection of biological data, study completion data, design of treatment provision and objective clinical assessment (see Supplementary table for details).Each study was assessed independently by two reviewers (AH and one of MA, PM, ÅW and DL), with any discrepancies resolved through discussion.Studies scoring below 1 were classified as having a high risk of bias (i.e.poor quality) while those scoring 1 or above were considered at low to moderate risk of bias (i.e.average to high quality).

Study selection
After removing duplicates, the search yielded 968 articles (Fig. 1).After initial screening of title and abstract, 64 articles were retrieved and assessed for inclusion.Evaluation of full-text articles yielded 27 reports of 26 studies that were included in this systematic review.One study (Bhattacharyya et al., 2019) was excluded as it reported on analyses performed on the same cohort as in the study by Gupta et al., 2016(Gupta et al., 2016) (the PGRN-AMPS cohort).Additionally, a third journal article reported results from the PGRN-AMPS cohort (Mah-moudianDehkordi et al., 2021); due to the inability to rule out potential overlapping results, this study was also excluded.One study was excluded (Perez et al., 1998) as it mainly presented data from three other included studies (Celada et al., 1992a;Celada et al., 1992b;Ortiz et al., 1993).Furthermore, another study was excluded as it did not report any definition of depression for the included patients (Mentre et al., 1998).

Study characteristics
A total of 26 studies and 27 reports were included in the review (the articles by Choi et al., 2021and Choi et al., 2022(Choi et al., 2022;Choi et al., 2021) present data from the same study and are reported together).Most of the studies were non-randomized and non-blinded (see Table 1).Two studies used the RDC criteria for major depression (Muck-Seler et al., 1991;Tyrer et al., 1990).All other studies used DSM criteria for major depressive disorder (MDD); however, a few studies also included patients with other depressive disorders, i.e. adjustment disorder with depressed mood (Alvarez et al., 1999), dysthymic disorder or depressive disorder not otherwise specified (NOS) (Choi et al., 2022;Choi et al., 2021), and "DSM depressive disorders" (not further specified) (Ortiz et al., 1993).
Most articles categorized patients as either responders or nonresponders at the end of the study based on a 50% or more reduction on a depression rating scale score.However, some authors used alternative definitions of treatment response, reported only on remission status or provided data on the percent change in depressive symptom scale scores (see Table 1).
Studies were heterogenous in terms of treatment provision, length of follow-up, choice of blood component and methods for analysis of 5-HT concentrations.Most studies only investigated the effect of pharmacological antidepressant treatment; two studies, however, also included electroconvulsive treatment (ECT) (Chappell et al., 2022;Lestra et al., 1998).One study only included adolescents (Goodnick et al., 2000) and one study included patients aged 17 and older (Choi et al., 2022;Choi et al., 2021); all other studies included adults aged 18 years or older.
As the reporting differed between the included studies, sample size is reported throughout the paper as the number of patients that had blood  There was only a nonsignificant trend for higher baseline platelet 5-HT content to relate to better response (r = 0.58, p = 0.10).
(continued on next page) A. Holck et al.There were no significant differences in platelet or plasma 5-HT between responders and nonresponders.
(continued on next page) A. Holck et al. samples drawn at baseline and started an antidepressant treatment unless otherwise specified.
Two studies investigated the effect of baseline platelet 5-HT on treatment effect of several different antidepressant treatments that were analyzed together.Lestra et al., 1998(Lestra et al., 1998) treated patients with paroxetine, viloxazine, moclobemide or ECT, and reported that improvement of depressive symptomatology (as measured by a decreased HDRS score) was negatively correlated with pretreatment 5-HT platelet content, i.e. lower baseline levels were associated with better treatment response.Muck-Seler et al., 1991(Muck-Seler et al., 1991) reported no difference in baseline platelet 5-HT levels between responders and non-responders treated with clovoxamine, amitriptyline, maprotiline, fluvoxamine or trazodone.
Three studies investigated the effect of baseline plasma 5-HT levels and response to non-SSRI antidepressant medications (phenelzine and brofaromine (Celada et al., 1992b), tianeptine (Ortiz et al., 1993) and desvenlafaxine XR (Sun et al., 2020)).None of these studies found an association between baseline plasma 5-HT levels and antidepressant treatment response.Chappell et al. 2002(Chappell et al., 2022) reported no significant association between clinical outcomes and baseline plasma 5-HT levels when evaluating the effect of various antidepressant medications and ECT together.

5-HT in serum
Six studies reported on the basal values of peripheral serum 5-HT in relation to treatment response in patients treated with antidepressant medications.Five of these investigated the effects of the SSRI treatments fluoxetine (Alvarez et al., 1999;Manoharan et al., 2016), sertraline (Zhu et al., 2013), paroxetine (Muck-Seler et al., 2002) and escitalopram (Xu et al., 2019).Only one of these studies reported an association between higher baseline 5-HT concentrations and better treatment response (Xu et al., 2019), whereas four studies reported no association (Alvarez et al., 1999;Manoharan et al., 2016;Muck-Seler et al., 2002;Zhu et al., 2013).Mück-Seler et al. 2002(Muck-Seler et al., 2002) also investigated the effect of the non-SSRI tianeptine in serum.They did not find an association between baseline 5-HT levels and antidepressant treatment response.Finally, Choi et al., 2021and Choi et al., 2022(Choi et al., 2022;Choi et al., 2021) found that higher baseline serum 5-HT concentrations were found in patients achieving remission after up to twelve weeks of treatment with various antidepressant medication separately or in combination.

5-HT in whole blood
Only one study investigated 5-HT levels in whole blood (Tyrer et al., 1990), and found no association between baseline 5-HT levels and outcome after fluoxetine treatment.

Statistical considerations
Due to between-studies heterogeneity with regards to study design, treatment provision and analyses, it was not possible to perform a metaanalysis.

Changes in peripheral 5-HT levels with treatment related to treatment outcome
Several studies reported how treatment-associated changes in 5-HT levels related to treatment outcome.Some studies found that larger decreases in 5-HT levels during treatment were associated with better treatment outcome.These associations were only found in studies reporting on SSRI treatment, specifically when 5-HT was analyzed in plasma after treatment with sertraline (Holck et al., 2019;Zhu et al., 2013) and citalopram (Gupta et al., 2016), and when 5-HT was analyzed in platelets (Dvojkovic et al., 2021) or plasma (Gupta et al., 2016) after treatment with escitalopram.
Contrary to these results, Celada et al., 1992b(Celada et al., 1992b) found that treatment with brofaromine and phenelzine significantly increased plasma 5-HT, more so in responders than in non-responders.

Quality assessment
The results of the quality assessment are presented in Table 2. Eleven out of the 26 studies were classified as having a high risk of bias (a quality assessment total score of <1; i.e. low quality).Low points were most frequently given for cohort formation, sample size and design of treatment provision.Fifteen studies were classified as having a moderate to low risk of bias (a quality assessment score of 1 to 7; i.e. average to high quality).
Fig. 2 shows all included studies grouped by result (higher 5-HT associated with better treatment outcome, lower 5-HT associated with better treatment outcome or no association between 5-HT levels and treatment outcome) in relation to quality score and cohort size.Note that studies that reported positive findings in only one of two or more treatment arms are included in the figure as positive results.Figs.3-5 only includes studies that reported on SSRI treatment separately from other treatments, and are divided into studies reporting results in platelets, plasma and serum, respectively.Since only one study reported on whole-blood, this is not represented by its own figure.

General discussion
To our knowledge this systematic review is the first attempt to summarize the relationship between baseline peripheral 5-HT levels and antidepressant treatment response.The included studies exhibited significant heterogeneity in methodology, type of intervention and outcomes, thus limiting the ability to draw firm conclusions.In all, we did not find robust support for an association between baseline 5-HT and treatment response, but results were mixed.Interestingly, no studies reporting on non-SSRI medications separately from SSRIs found an association between baseline 5-HT levels and antidepressant treatment response.These results suggest that SSRIs may have distinct relationships to peripheral 5-HT compared to other types of antidepressant medication.Despite mixed results, this indicates the potential value of further research into peripheral 5-HT levels as a possible biomarker of   specifically SSRI treatment outcome, as higher baseline 5-HT was associated with better response in several SSRI studies, including several larger and higher-quality studies.Moreover, the results from the present study confirm previous findings that SSRI treatment reduces 5-HT levels regardless of treatment response.Finally, we found no evidence of an association between greater treatment-associated changes in 5-HT levels and a superior clinical response.
The acute effect of SSRI binding to the serotonin transporter (SERT) in the central nervous system is a decrease of transporter affinity for 5-HT, resulting in an accumulation of 5-HT in the synaptic cleft (Stahl, 1998).While theories on the delayed efficacy of SSRIs suggest neuroadaptive changes such as desensitization of 5-HT1A autoreceptors, leading to increased presynaptic output of 5-HT (Stahl, 1998), these effects are not unique to SSRIs.Even though SSRIs are presumed to primarily act on the serotonergic system, effects on many other systems have also been proposed, such as the glucocorticoid system (Feighner, 1999) and down-stream effects through compounds such as brainderived neurotrophic factor (BDNF) and glutamate leading to neuroplastic effects (Harmer et al., 2017;Racagni and Popoli, 2008).The comprehensive mechanisms underlying SSRI effects and their association with peripheral 5-HT levels remain complex, with insights from tryptophan depletion (Jauhar et al., 2023;Ruhe et al., 2007) and the influence of inflammation on tryptophan metabolism (Chen and Guillemin, 2009;Pan et al., 2022;Zhou et al., 2022) highlighting potential avenues; yet, a detailed mechanistic discussion exceeds the scope of this review.
Our investigation into the association between changes in peripheral 5-HT levels and treatment outcome revealed mixed findings.While some studies suggested that a larger decrease in 5-HT levels during treatment was associated with a better clinical response to SSRI treatment (Dvojkovic et al., 2021;Gupta et al., 2016;Holck et al., 2019;Zhu et al., 2013), several other studies did not find this association (Figueras et al., 1999;Manoharan et al., 2016;Narayan et al., 1998;Sun et al., 2020).In the only study that reported any association between changes in 5-HT levels and treatment outcome for non-SSRIs, an increase in 5-HT levels was associated with better treatment response (Celada et al., 1992b).This aligns with expectations, as the patients were treated with monoamine oxidase (MAO) inhibitors brofaromine and phenelzine, medications which inhibit 5-HT metabolism and are known to increase brain 5-HT levels (Stahl and Felker, 2008).The review search strategy was not developed for this specific outcome, which limits the comprehensiveness of these findings.
There was no clear pattern with regards to how 5-HT analysis in different blood components impacted the association between baseline concentration or change in 5-HT and treatment outcome.Platelets were the most studied blood component as reported in 16 studies, followed by ten studies with plasma 5-HT, six with serum 5-HT and a single study with whole blood 5-HT.Results varied greatly in studies examining the effect of SSRI treatment and baseline 5-HT in platelets (Castrogiovanni et al., 2003;Celada et al., 1992a;Dvojkovic et al., 2021;Figueras et al., 1999;Goodnick et al., 1995;Goodnick et al., 1997;Ko et al., 1997;Kuhs et al., 1992;Lestra et al., 1998;Muck-Seler et al., 2002;Narayan et al., 1998;Pivac et al., 2003) and plasma (Alvarez et al., 1999;Castrogiovanni et al., 2003;Celada et al., 1992a;Figueras et al., 1999;Gupta et al., 2016;Holck et al., 2019;Sun et al., 2020).In serum, four out of five studies reported no association between baseline 5-HT and treatment outcome (Alvarez et al., 1999;Manoharan et al., 2016;Muck-Seler et al., 2002;Zhu et al., 2013); however, the fifth and the largest study did find an association between higher baseline 5-HT and better treatment response (Xu et al., 2019).Whole blood 5-HT was analyzed in only one study (Tyrer et al., 1990).Previous studies have suggested that results from studies of 5-HT levels in platelet-poor plasma are highly discrepant, most likely due to the difficulty of measuring the extremely low concentrations that may also be contaminated from platelet-derived 5-HT (Brand and Anderson, 2011) and its rapid metabolization in the circulation (Szeitz and Bandiera, 2018); however, the results from the  studies that measured 5-HT in plasma did not stand out in terms of outcome.That being said, most studies did not report exact 5-HT values.In future studies it would be highly relevant to further investigate the differences between different blood components, as the different peripheral 5-HT compartments react to distinct homeostatic signals and are unlikely to be interchangeable as predictors of treatment response.Additionally, the different assay methods, sample handling and storage conditions may impact the results and lead to increased heterogeneity between study results.
Regarding diagnostic criteria, all except two of the included studies used DSM criteria.Most studies included patients with MDD; a few studies expanded the population to include other depressive disorders.However, the MDD diagnosis in itself is well-known to be heterogenous in terms of both symptomatology and pathophysiology (Goldberg, 2011;Zimmerman et al., 2015).The patients included in the different cohorts are likely to differ substantially depending on a variety of factors such as in-or outpatient status, recent suicide attempt, and comorbidities.The authors of one of the studies included in this review concluded that "unless a robust classification of MDD subtypes with their respective underlying pathways is charted, finding a clinically relevant treatment response biomarker may remain elusive" (Manoharan et al., 2016); this heterogeneity of diagnosis may well be a contributing factor to the inconsistent results of previous studies.That two out of the 26 studies diagnosed patients according to the RDC (Muck-Seler et al., 1991;Tyrer et al., 1990) instead of DSM criteria is probably less likely to have affected the results as the diagnostic criteria are very similar.

Limitations
Several limitations influence the results of the included studies and consequently this systematic review.
Many of the included studies were very small and are likely to have been underpowered.Only a minority of the studies were randomized controlled trials.Furthermore, several of the included studies were exploratory in nature and examined several different possible predictors of antidepressant treatment outcome.These studies generally did not adjust their results for multiple comparisons, increasing the risk for false positive results.
Most of the included studies reported a washout period of medications before study enrollment; however, even when reported, this period varied greatly from just three days (Kuhs et al., 1992) to several months (Alvarez et al., 1999;Ko et al., 1997).As our results show that SSRI treatment lowers peripheral 5-HT levels, recent treatment with similar substances may impact the baseline values.
While some studies controlled for confounding factors such as age, gender, comorbidities and medication regimens, others did not.This overall heterogeneity of both study design and outcome reporting made it impossible to perform a meta-analysis thus limiting the interpretation of the results.
Another important limitation is that this review combines results from studies reporting 5-HT levels in different peripheral blood compartments.While platelets were the most studied blood components, these studies were generally older and with lower quality than studies reporting on 5-HT levels in plasma or serum.Plasma, however, may be a less appropriate blood component for 5-HT analysis (Brand and Anderson, 2011).
While interpreting the results of this review, it is important to recognize that peripheral 5-HT levels do not fully mirror central 5-HT levels.Platelets in particular have been suggested as comparable to neurons in serotonergic actions, as they are similar in terms of 5-HT uptake by SERT, storage, metabolism and release mechanisms (Zhuang et al., 2018).There have been reports of strong correlations between cerebrospinal fluid (CSF) and platelet 5-HT levels (Audhya et al., 2012), as well as moderately strong correlations between CSF and plasma 5-HT levels (Audhya et al., 2012;Gao et al., 2008).Significant and positive correlations between levels of the 5-HT metabolite 5-hydroxyindolacetic acid (5-HIAA) in CSF and plasma have also been reported within individuals (Strawn et al., 2002).These results suggest that peripheral 5-HT levels may in fact, at least partly, inform about central serotonergic mechanisms; however, with regards to serum and whole-blood, current knowledge is sparse.Although understanding any relationships between peripheral 5-HT and treatment response would be valuable, future studies assessing baseline 5-HT concentration, as well as changes in 5-HT with treatment, as predictors for SSRI treatment response would be of value regardless of the underlying biological changes responsible for the effect.

Conclusion
In conclusion, the findings from this systematic review indicates that for depressive disorders, there appears to be no association between baseline peripheral 5-HT levels and treatment outcome in non-SSRI antidepressant treatment.Results regarding SSRIs are inconsistent; however, several larger studies and studies with lower risk of bias showed that higher baseline 5-HT was associated with greater antidepressant response to SSRIs, encouraging future research in this area.
Our results confirm previous findings that SSRI treatment is associated with a decrease in peripheral 5-HT levels; however, we found no clear evidence that greater decrease in 5-HT is associated with improved treatment outcome, although results were inconsistent.
To enable meaningful future comparisons and meta-analyses, robust, large-scale, randomized, double-blinded studies with well-defined cohorts are imperative.Additionally, future research should attempt to replicate higher-quality existing studies to reinforce and validate existing studies.
Should future research support an association between peripheral baseline 5-HT levels and SSRI treatment outcome in depressive disorders, standardizing the choice of blood component and assay methods to establish cut-off values for 5-HT levels predictive of SSRI responsiveness would be an important next step.

Sources of direct funding, support, or sponsorship
A Holck was supported by the Southern Health Care Region, Sweden.Å Westrin was supported by Swedish governmental funding of clinical research (ALF).D Lindqvist was funded by the Swedish Research Council and Swedish governmental funding of clinical research (ALF).OM Wolkowitz was funded by a gift from the Tinberg Family.

Role of the funding source
The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Amanda Holck reports financial support was provided by Southern Health Care Region, Sweden.Asa Westrin reports financial support was provided by Swedish governmental funding of clinical research (ALF).Daniel Lindqvist reports financial support was provided by Swedish governmental funding of clinical research (ALF).Daniel Lindqvist reports financial support was provided by Swedish Research Council.Owen M. Wolkowitz reports financial support was provided by Tinberg Fanily (gift).Pouya Movahed reports a relationship with H Lundbeck AB that includes: speaking and lecture fees.Marie Asp reports a relationship with H Lundbeck AB that includes: speaking and lecture fees.Daniel Lindqvist reports a relationship with H Lundbeck AB that includes: speaking and lecture fees.Daniel Lindqvist reports a relationship with Janssen Cilag AB that includes: speaking and lecture fees.If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. Quality score and cohort size of the 26 studies in relation to the reported associations between baseline 5-HT levels and better treatment outcome in all included studies.The sample size is represented by the size of the bubbles ranging from 10 to 1094.Some studies with identical quality score overlap resulting in bubbles of darker colours.5-HT: 5-hydroxytryptamine (serotonin).

Fig. 3 .
Fig. 3. Quality score and cohort size of the twelve studies that reported on SSRI treatment separately from other treatments in relation to the reported associations between baseline 5-HT levels and better treatment outcome in platelets.The sample size is represented by the size of the bubbles ranging from 11 to 123.Some studies with identical quality score overlap resulting in bubbles of darker colours.5-HT: 5-hydroxytryptamine (serotonin).

Fig. 4 .
Fig. 4. Quality score and cohort size of the seven studies that reported on SSRI treatment separately from other treatments in relation to the reported associations between baseline 5-HT levels and better treatment outcome in plasma.The sample size is represented by the size of the bubbles ranging from 10 to 306.Some studies with identical quality score overlap resulting in bubbles of darker colours.5-HT: 5-hydroxytryptamine (serotonin).

Fig. 5 .
Fig. 5. Quality score and cohort size of the five studies that reported on SSRI treatment separately from other treatments in relation to the reported associations between baseline 5-HT levels and better treatment outcome in serum.The sample size is represented by the size of the bubbles ranging from 10 to 306.Some studies with identical quality score overlap resulting in bubbles of darker colours.5-HT: 5-hydroxytryptamine (serotonin).

Table 1
Characteristics of included studies.

Table 2
Result of the quality assessment of the included studies.