Psycho-Pharmacomicrobiomics: A Systematic Review and Meta-Analysis

,

About 1 in 4 human-targeted medications can inhibit the growth of gut bacteria, and a large proportion of these medications are antipsychotics (1,2).Medication-induced changes in the gut microbiome can significantly affect their efficacy and tolerability (3).A seminal example is metformin, which improves glucose tolerance by modifying the gut microbiome (4).On the other hand, gut bacteria can modify medication metabolism, which may alter their efficacy and tolerability (5).Human gut bacteria can convert promedications into active compounds (6) or inactivate medications (7), influence their enterohepatic recirculation via b-D-glucuronidases (8), and affect their absorption via secondary bile acids and blockage of intestinal P-glycoproteins (9).Bacteria can also directly influence the tolerability of medications by generating toxic metabolites via bacterial enzymes (10).
The study of the interplay between bacteria and medications has been described as pharmacomicrobiomics (11).Due to the modifiable nature of the gut microbiome, findings can also help identify modifiable targets of interventions to improve the efficacy and tolerability of existing medications.
Pharmacomicrobiomics applied to the field of psychiatry (psycho-pharmacomicrobiomics) may help address key unmet needs, such as the high interindividual variability in treatment response and the high burden of side effects of psychotropic medications (12).A nationwide population study showed that the use of antibiotics, which are strong modifiers of the gut microbiome, was associated with increased prescriptions of psychotropics (antipsychotics, mood stabilizers, and antidepressants) and hospitalizations for mental health disorders (13).These findings, which provide indirect evidence of the association between alterations in gut microbiome and reduced efficacy of psychotropics, are supported by findings from an animal study that showed a direct relationship between changes in gut bacteria and reduced bioavailability of olanzapine (14).Therefore, studies on the interplay between psychotropics and the gut microbiome in clinical populations are emerging.However, findings from individual microbiome studies are often hard to interpret due to inconsistencies in methodological approaches and reporting bias.For these reasons, we conducted a systematic review and meta-analysis in which we summarized the evidence from clinical studies that investigated the bidirectional interplay between the gut microbiome and psychotropic medications.We addressed 2 key questions: 1) Do psychotropic medications modify the gut microbiome? and 2) Does the gut microbiome affect the efficacy and tolerability of psychotropic medications?
Finally, we systematically appraised the quality of the included studies against the newly validated STORMS (Strengthening The Organization and Reporting of Microbiome Studies) checklist (15) and provide directions for future investigations in the field.

METHODS AND MATERIALS
We followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting guidelines (16) (see the Supplement for the checklist) and preregistered our protocol (https://osf.io/zde5y/).
A search was conducted from inception to November 2022 in Web of Science and PubMed.
Studies that did not meet criteria for inclusion were excluded (Figure 1).Search process and data extraction were conducted independently by 2 authors, and disagreements were resolved by consensus.

Type of Studies
Inclusion criteria were longitudinal studies in clinical populations that investigated changes in gut microbiome profiles before and after starting a treatment with a psychotropic medication(s) and cross-sectional studies that compared gut microbiome profiles in patients treated with psychotropic drugs versus untreated patients.
There was no restriction on the underlying clinical conditions being treated with psychotropic medications (antidepressants, antipsychotics, mood stabilizers, or their combination) or on the age, gender, or body mass index of participants.Only original studies written in English were included.
Exclusion criteria were studies that collected gut microbiome data in medicated patients and healthy control participants because differences in the characteristics of gut microbiome communities reported in these studies may be related to illness rather than to medications.For the same reason, we excluded studies of participants with comorbid gastrointestinal disorders (e.g., major depressive disorder with inflammatory bowel disorders) and healthy volunteers only.Data reported in conference abstracts and unpublished literature were also excluded.

Outcomes
Primary outcomes were differences in gut microbiome [i.e., diversity, taxonomy, and functionality (17)] before and after treatment with psychotropic medications (in longitudinal studies) and in medicated compared with unmedicated patients (in cross-sectional studies).
Secondary outcomes were associations between gut microbiome and measures of treatment response and tolerability across the included studies.Where sufficient data were available, differences in gut microbiome profiles in treatment responders (as defined by individual studies) were compared with nonresponders.

Data Synthesis
Random effect meta-analyses were performed for alpha diversity metrics; summary statistics included standardized mean differences (SMDs), 95% CI, and the Higgins I 2 .
Synthesis of beta diversity data followed a different, yet validated approach (18).In brief, we extracted individual patient data at the coordinates of the first 2 axes of beta diversity metrics from each study.We then performed permutational multivariate analysis of variance with the vegan adonis function in R on the Euclidian distance matrix of the extracted data; summary statistics included R 2 , F, and p values.
Quantitative analyses were conducted separately for longitudinal and cross-sectional studies.
Preplanned subgroup analyses were based on variables with a known impact on the gut microbiome.Microbial taxonomic and functional findings were summarized following previously validated methods (19).

Risk of Bias/Quality of Included Studies
Risk of bias was assessed independently by 2 reviewers using an adapted version of the STORMS checklist (15) and visualized with the robvis tool (20).
See Tables 1 and 2 for a qualitative summary of the results.
Five studies reported data on gut microbiome alpha diversity before and after treatment with antipsychotics (28,30,32,33,38).All studies reported the Shannon index as a measure of alpha diversity.One of these studies did not report analyzable data (38); therefore, it was excluded from the analyses.The pooled estimate showed a significant increase in gut microbiome alpha diversity after treatment with antipsychotic medications (n = 4 studies; SMD: 0.12; 95% CI: 0.01-0.23;p = .04;I 2 : 14%) (Figure 2).
Only 1 study included obese participants (30), which precluded subgroup analyses based on obesity.None of the studies included in the quantitative synthesis on alpha diversity included children, reported analyzable data on gender, or used polypharmacy, and all used the same sequencing method; follow-up timing ranged from 1 to 6 months.
Four longitudinal studies reported data on beta diversity before and after treatment with antipsychotics (28,30,33,38).When individual data from these studies were pooled, no significant difference was found in beta diversity before versus after treatment (F = 0.38; R 2 = 0.01; p = .68)(Supplemental eFigure 1).
All 8 longitudinal studies reported data on gut microbiome taxonomy; all but 1 study (30) reported significant changes at phylum, class, family, and genus levels.Across studies, the most consistent findings were of increased Gammaproteobacteria at the class level and increased Bifidobacteria, Lactobacillus, and Klebsiella at the genus level after treatment with antipsychotics (see also Supplemental eTable 1).
Three studies reported data on microbiome functional analysis following treatment with antipsychotics (28,30,33).Two of 3 studies reported significant changes across 12 different microbiome functional pathways, including those involved in the metabolisms of short-chain free fatty acids (28,33) (see Supplemental eTable 1).
All 5 studies reported data on alpha diversity.The pooled effect size showed no differences in alpha diversity between antipsychotic-treated and antipsychotic-free patients (SMD: 20.08; 95% CI: 20.68 to 0.53; p = .80)with evidence of high heterogeneity (I 2 : 83.98%).
Two studies reported data on gender differences in alpha diversity metrics between antipsychotic-treated and -untreated patients (35,36).These 2 studies were also the only ones that included obese participants in the study population.The pooled estimate showed a significant reduction in alpha diversity in female, but not male, patients (SMD: 20.68; 95% CI: 21.31 to 20.04; p = .04;I 2 : 37.61%).
All studies included in the quantitative synthesis on alpha diversity in cross-sectional studies used the same sequencing a This study was not reported in the longitudinal (within-comparison) table, and longitudinal data were not used to produce meta-analytic estimates because the authors compared differences in alpha diversity between treatment groups (e.g., changes in microbial diversity after antipsychotic treatment vs. changes in microbial diversity after antidepressant treatment) rather than pre-to-post differences within the same treatment group.b Defined on "retrospective assessment of longitudinally collected information."method, none were conducted with children, all but 1 (37) used polypharmacy, and all included participants in long-term treatment with antipsychotics (i.e., .6 months).Three cross-sectional studies reported data on beta diversity in antipsychotic-treated versus antipsychotic-free patients (27,35,38).When individual data from these studies were pooled together, we found a significant difference in beta diversity between antipsychotic-treated and antipsychotic-free patients (F = 3.31; R 2 = 0.02; p = .03)(Supplemental eFigure 2).Across these 3 studies, only 1 study included obese participants in their study population (35), 1 was conducted with children and adolescents ( 27), all used polypharmacy, and all included participants who were receiving long-term treatment with antipsychotics.
All cross-sectional studies except 1 (38) reported data on gut microbiome taxonomy in antipsychotic-treated versus antipsychotic-free participants; all reported significant between-groups differences at phylum, class, family, and genus levels.However, specific taxonomic findings were not consistent across studies (Supplemental eTable 2).
Three studies reported data on microbiome functional analysis in antipsychotic-treated versus antipsychotic-free participants (27,37,38).Two of 3 studies reported significant differences in the expression of microbial functional pathways; both studies reported an increase in tryptophan metabolism microbial pathways in antipsychotic-treated patients (27,38) (Supplemental eTable 2).
Association With Clinical Outcomes.Three longitudinal studies reported data on the association between gut microbiome features and treatment response to antipsychotics (28,30,32).
Two of 3 studies showed that gut microbiome features at baseline, including highly represented species belonging to the Ruminococcaceae and Lachnospiraceae families, were associated with a subsequent therapeutic response to antipsychotics (28,32) (Table 1).The only study that reported negative findings was conducted in a cohort of multiepisode chronic patients with schizophrenia, which included obese participants (30).
Only 1 (32) of 3 (28,30,32) studies reported that gut microbiome changes following antipsychotic treatment were associated with treatment response.This study investigated risperidone and was conducted with a large cohort of adolescents and young adults during their first psychotic episode (32).
No studies on antipsychotics reported analyzable data to quantitatively compare differences in gut microbiome features in treatment responders versus nonresponders.
Three longitudinal studies reported data on the association between baseline gut microbiome features and tolerability of antipsychotics (30,31,34) Two of these 3 studies reported that both baseline and changes in gut microbiome features were associated with adverse metabolic outcomes (i.e., higher blood levels of lowdensity lipoproteins and triglycerides; increased weight gain) following treatment with antipsychotics (31,34); the only study that reported negative findings was conducted in a cohort of multiepisode chronic patients with schizophrenia, which included obese participants (30).
None of the studies included in the quantitative analyses included obese participants, none was conducted in children or adolescents, none reported data on gender differences, none used polypharmacy, and all used the same sequencing method.Follow-up timing ranged from 1.5 to 6 months.
Two studies reported data on microbiome functional analysis following treatment with antidepressants (22,25).One reported significant changes in microbial pathways related to biosynthesis of secondary metabolites, while the other found no changes following antidepressant treatment (28,33).
None reported data on alpha or beta diversity.One study reported data on gut microbiome taxonomy and showed significant differences in the abundance of a number of bacterial species in antidepressant-treated versus antidepressant-free participants (39) (Table 2).One study that reported data on gut microbiome functionality found no differences between antidepressant-treated versus antidepressant-free participants (38).
Association With Clinical Outcomes.Two longitudinal studies reported data on baseline alpha diversity and subsequent treatment response to antidepressants, and both studies investigated escitalopram (21,23).The pooled estimate showed that patients who remitted from depression following treatment with escitalopram had higher alpha diversity at baseline compared with patients who did not remit (SMD: 2.45, 95% CI: 0.50-4.40,p , .001).
Three longitudinal studies reported taxonomic data on the association between gut microbiome features and treatment response to antidepressants (21,23,24).Two of 3 studies showed that baseline and changes in gut microbiome taxonomic features were associated with treatment response to antidepressants (21,24) (Table 1).
These findings were consistent with the only crosssectional study that investigated differences in gut microbiome taxonomy in antidepressant responders versus nonresponders (39).

Methodological Considerations and Risk of Bias
None of the included studies was at low risk of bias (Supplement).Main limitations included the lack of supporting metatranscriptomics or metabolomics data and differences between groups in terms of polypharmacy and other confounders.

DISCUSSION
To our knowledge, this is the first systematic review and metaanalysis of the effects of psychotropic medication on the gut microbiome in clinical populations.
The main findings were 1) treatment with antipsychotics and antidepressants were associated with significant changes in gut microbiome diversity indices, taxonomy, and functionality; 2) these gut microbiome parameters at baseline were associated with subsequent therapeutic response to both antidepressants and antipsychotics; 3) there were gender-specific differences in the gut microbiome associated with exposure to antipsychotics wherein female participants had reduced alpha diversity compared with male participants; and 4) most included studies present significant methodological limitations.

Antipsychotics
Our pooled estimates on longitudinal studies suggest that treatment with antipsychotics increases alpha diversity in patients.Low alpha diversity has been associated with detrimental health outcomes, including mood and psychotic disorders (40)(41)(42).A recent meta-analysis showed that patients with psychiatric disorders had lower alpha diversity compared with healthy control participants (19).One hypothesis is that antipsychotics may increase and normalize the diversity of the gut flora, making it closer to the one of healthy control participants, with downstream potential benefits for patients (37,43).The qualitative summary on antipsychotic-related changes in taxonomy and functionality of the gut microbiome seem to point toward a similar direction.Antipsychotic-related gut microbiome changes included an increase in butyrateproducing bacteria, such as Bifidobacteria and Lactobacillus, and in the expression of butyrate metabolic pathways, which have known procognitive and anti-inflammatory actions (19).However, to date, only a few studies have directly investigated the relevance of these antipsychotic-related gut microbiome changes for treatment outcomes.Those that reported data on drug-naïve patients with early psychosis suggest that gut microbiome changes may be clinically informative (31,32), consistent with findings from a recent study showing an association between increased serum levels of butyrate and treatment response to risperidone in drug-naïve patients with first-episode psychosis (44).
Evidence from in vitro and animal studies suggests that some antipsychotics can directly inhibit or enhance the growth of gut bacteria (1,2).This has led some authors to hypothesize that antipsychotics may act via the gut microbiome (45).However, the observed antipsychotic-related increase in alpha diversity and in the butyrate-producing activity of gut bacteria may simply be a consequence of an improvement in illness severity.A better symptom profile (e.g., less paranoia) may increase patients' sociability and habits (e.g., more interpersonal interactions, more active lifestyle, increase exposure to outdoor spaces), which in turn could increase alpha diversity (46).Future studies using translational approaches, such as gnotobiotic and/or in vitro studies (47), are needed to clarify the direction of causality.
Although not significant, the pooled estimate from crosssectional studies showed reduced alpha diversity in patients who were treated with antipsychotics compared with untreated patients, a finding that went in an opposite direction from the one observed in longitudinal studies.Many of the side effects of antipsychotics are metabolic and include weight gain (48), which is known to reduce gut microbial alpha diversity (49).Because all cross-sectional studies on antipsychotics were conducted with chronically treated patients, it is possible that the observed reduction in alpha diversity may mirror the longer-term consequences of antipsychotic treatment wherein chronicity-related mechanisms (including but not limited to weight gain) may counteract the initial increase in alpha diversity reported in longitudinal studies.In support of this hypothesis, the most significant increase in alpha diversity following antipsychotic treatment was observed in drug-naïve patients with first-episode psychosis (32), while negative findings were reported by the only longitudinal study that Microbiome and Psychiatric Medications included multiepisode chronic patients with schizophrenia and comorbid obesity (30).A longer exposure to medications may also be responsible for the observed significant differences in gut microbiome structure (beta diversity) in cross-sectional, but not in longitudinal studies.
Notably, when analyses on cross-sectional studies were restricted to female participants only, we found a significant reduction in alpha diversity compared with untreated female control participants.Compared with male patients, female patients are at greater risk of developing metabolic side effects from antipsychotics (50,51).Some authors suggest that this may be due to use of excessive doses in female patients or to hormonal factors (52).Because the gut microbiome has been causally involved in the development of metabolic disorders (53), future studies should investigate whether the observed gender differences in alpha diversity associated with antipsychotic treatment may drive the metabolic side effects that are experienced by patients.

Antidepressants
Pooled data on longitudinal studies on antidepressants showed a nonsignificant increase in alpha diversity and a significant difference in beta diversity following treatment.
The lack of significance in the pooled effect size for alpha diversity may be due to the low number of participants across studies (one-third of those available for antipsychotics) rather than a complete lack of effect because beta diversity metrics suggest that antidepressants are significantly associated with changes in gut microbiome structure.The majority of included studies suggested that antidepressant-related changes in gut microbiome may be clinically informative.The taxonomic summary showed an increase in Christensenellaceae following treatment with antidepressants.Low levels of this bacterium have been found in patients with mood disorders and associated with more severe depressive symptoms (54,55); novel probiotics formulations containing Christensenellaceae are currently being tested as antidepressants (56).These findings suggest that antidepressants may act via the gut microbiome.However, as discussed for antipsychotics, it is also possible that antidepressant-related changes in gut microbiome may just represent an indirect consequence of improvement in symptoms.
Our analysis of gut microbiome data in treatment responders compared with nonresponders showed that lower alpha diversity at baseline was associated with a lack of response to subsequent treatment with antidepressants.The qualitative summary on longitudinal studies on antipsychotics also corroborated this finding, further suggesting that baseline gut microbiome features may affect the tolerability of this class of psychotropics.
Gut bacteria have the potential to significantly affect the bioavailability of drugs (57).A recent study reported that modifications in the gut microbiome can double the bioavailability of olanzapine (14).This could occur via direct and indirect mechanisms.Indirect mechanisms include the ability of gut bacteria to modify the activity of human cytochromes (58) and the enterohepatic recycling of drugs (8); direct mechanisms include biotransformation (59) and bioaccumulation (60).Recently, it has been shown that the expression of gut bacterial tyrosine decarboxylase can reduce the bioavailability of levodopa, a drug used for the treatment of Parkinson's disease, and increase the level of m-tyramine, with direct consequences on its efficacy and tolerability (59).
Changes in gut microbiome, such as low alpha diversity, have been associated with an increased chronic aspecific proinflammatory status (i.e., increased C-reactive protein, cytokine levels) (41,61), which may underlie certain forms of treatment resistance in both mood and psychotic disorders (62,63).However, the precise mechanisms by which inflammation should drive treatment resistance is unknown, and treatment stratification strategies based on C-reactive protein/ cytokines have not proven effective so far (64,65).One possibility is that patients with this increased chronic proinflammatory status may just have an unfavorable gut microbiome composition (61), which may drive-with more specific mechanisms-certain aspects of treatment resistance.
None of the included studies was at low risk of bias.A main issue is that findings from individual studies often did not account for important variables such as metabolic changes, diet, obesity, and gender.This limits the robustness of the conclusions that can be drawn.

Limitations
Data from antipsychotics and antidepressants were pooled together based on their similarities on central brain effects.However, it is likely that within the same category, psychotropics may differ greatly in terms of effects on the gut microbiome.At the same time, it is possible that gut bacteria may contribute to the metabolism of only a few compounds rather than the whole class of psychotropics.None of the included studies investigated mood stabilizers; therefore, we could not report any data on this class of medications.

Conclusions and Future Directions
This systematic review and meta-analysis showed that psychotropic medications are associated with altered gut microbiome in patients with psychotic and mood disorders and that the interaction between psychotropics and gut bacteria can potentially be informative for efficacy and tolerability outcomes in psychiatric illness.
Because about one-third of patients do not respond to antipsychotics or antidepressants (66), clarifying whether and how the gut microbiome contributes to the efficacy of these compounds is of utmost importance.For antipsychotics, this may also contribute to their poor metabolic tolerability, which is a key unmet need in psychotic disorders.
Studies on treatment-naïve patients would be helpful in clarifying the extent to which the observed changes in the microbiome are related either to the disease status or to the direct effects of medications.Regardless of the cause of the microbiome change (disease, lifestyle, or medication), its predictive value as a biomarker remains valid.Therefore, findings from treated clinical populations would still be highly valuable (also because they are more representative of real-world scenarios).In this case, direct drug-microbiome interactions can be explored using in vitro models, where individual patients' microbes are incubated with the medication of interest.Gut microbiome taxonomic data should always be enriched with Microbiome and Psychiatric Medications Biological Psychiatry April 1, 2024; 95:611-628 www.sobp.org/journalfunctional and metabolomics analysis (67) because same taxa could differ significantly in terms of their physiological role and, at the same time, distinct taxa could fulfill the same functional role (17).Standardized procedures, such as the STORMS guidelines (15), should be adopted to increase reproducibility of findings.
Ultimately, a multidisciplinary approach that incorporates well-powered clinical studies, in vitro experiments, and gnotobiotic animal models is essential to provide answers that can be translated into real-world clinical applications.Translational studies will help elucidate the issue of reverse causality, such as whether psychotropic medications modify the microbiome or whether the observed changes are a result of altered lifestyle or the disease itself.These studies will also help dissect the contribution of the host-microbiome interaction to treatment outcomes.
Finally, most, if not all, microbiome-targeted interventions (e.g., pre/probiotics) tested in clinical trials for psychosis or depression have been developed with the intention of targeting the underlying pathophysiology of these syndromes.However, this approach has proven unsuccessful in psychosis (42) and only partially successful in mood disorders.Therefore, the next generation of microbiome-targeted interventions should focus on enhancing the effectiveness and/or improving the tolerability of antipsychotics and antidepressants.

Figure 1 .
Figure 1.PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart.

Table 2 .
Qualitative Summary of Cross-sectional (Between-Comparisons) Studies