This study constitutes a systematic review with integrated meta-analysis, conducted in strict accordance with the PRISMA 2020 guidelines (15). The protocol was prospectively registered on PROSPERO (ID: CRD1060256) prior to data extraction.

A comprehensive search was performed across four electronic databases (PubMed, Web of Science, Embase, PsycINFO) from January 2014 to March 2025, using a three-tiered strategy:

Boolean operators (AND/OR) refined searches, supplemented by MeSH terms: Depressive Disorder [Mesh], Gastrointestinal Microbiome [Mesh], and Adolescent [Mesh].

Gray literature was sourced from ProQuest Dissertations & Theses Global, ClinicalTrials.gov, and ISRCTN Registry to mitigate publication bias. Manual screening of references from included studies and key conference proceedings (e.g., International Society for Microbiota) ensured coverage.

Inclusion: (1) Original studies investigating gut microbiome alterations/interventions in adolescent depression (mean age ≤19 years); (2) human trials (RCTs, cohorts, case–control), animal models, or meta-analyses; (3) English-language publications with empirical data.

Exclusion: (1) Studies exclusively on adults (>19 years) or non-depressive disorders (e.g., anxiety alone); (2) non-microbiome mechanistic studies (e.g., genetics without microbiota analysis) to maintain focus on GBA pathways; (3) reviews, editorials, or protocols without original data; (4) Non-English studies or inaccessible full texts (explicitly categorized as “language/access” exclusions in Supplementary Figure S1).

Two independent reviewers screened titles/abstracts and full texts using Covidence® software (Veritas Health Innovation). Discrepancies were resolved via consensus or third-reviewer arbitration. The PRISMA flow diagram (Supplementary Figure S1) details the selection process:

Data were extracted using a standardized template: (1) study design, sample size, participant demographics; (2) microbiome metrics (α-diversity, taxa abundance); (3) clinical/behavioral outcomes (e.g., HAM-D scores); (4) intervention details (strain, dosage, duration).

Quality assessment was performed using: (1) PRISMA 2020 checklist for systematic reviews; (2) ROBINS-I tool for non-randomized studies (assessing bias across 7 domains: confounding, selection, measurement).

Studies were rated as low, moderate, or high risk of bias. Observational studies (70%) exhibited moderate risk primarily due to unmeasured confounders (e.g., diet).

A random-effects model (RevMan 5.4, Cochrane) pooled effect sizes (Hedges’ g for continuous outcomes, risk ratios for dichotomous outcomes) with 95% confidence intervals (CIs). Heterogeneity was quantified via I² statistics (I² > 50% = substantial). Subgroup analyses examined: (1) age (early [10–14 years] vs. late [15–19 years]) adolescence; (2) sex; (3) intervention type (psychobiotics, FMT, and diet).

Sensitivity analyses excluded studies with high risk of bias.

Meta-analysis of 15 studies (n = 1,200 adolescents) revealed that depressed adolescents exhibited significantly reduced gut microbiota α-diversity vs. healthy controls (Shannon index SMD = −0.92; 95% CI: −1.24, −0.60; I² = 68%; p < 0.001; Figure 2A). Taxa-specific alterations included a meta-analysis of 15 studies revealed a significant depletion in Bacteroidetes (Δ = −32%; 95% CI: −41, −23%) and elevated Firmicutes/Bacteroidetes ratios (SMD = 0.85; 95% CI: 0.42, 1.28). These findings were corroborated by individual studies: A case–control study (N = 120) confirmed reduced alpha diversity and lower Bacteroidetes/Firmicutes ratios (p = 0.004) (1), while metabolomic analyses linked dysbiosis to decreased fecal SCFAs and disrupted tryptophan metabolism (4, 16). Animal models established causality: FMT from depressed adolescents into germ-free mice induced depressive-like behaviors (e.g., reduced sucrose preference; p < 0.05) and neuroinflammation (hippocampal IL-6↑ 45%, TNF-α↑38%) (17). Caution is warranted due to limited preclinical sample sizes (e.g., N = 20).

Neuroinflammation: Gut dysbiosis activates TLR4/NF-κB signaling in the prefrontal cortex, promoting astrocyte reactivity and IL-1β release (18). Certain Clostridium species (e.g., C. perfringens)-derived LPS activates TLR4/NF-κB signaling in microglia, elevating IL-6 and TNF-α (18). Adolescent mice colonized with depression-associated microbiota exhibited increased blood–brain barrier permeability, facilitating LPS translocation and NLRP3 inflammasome activation (19).

Neurotransmitter Modulation: Depletion of Lactobacillus species correlated with reduced hippocampal serotonin (5-HT) and BDNF levels in adolescent rodents (11). Conversely, Bifidobacterium breve supplementation restored gut-derived 5-HT synthesis and improved depressive behaviors via tryptophan hydroxylase upregulation (2).

Intestinal Barrier Dysfunction: Elevated serum zonulin and fecal calprotectin levels in depressed adolescents indicated compromised gut barrier integrity, which correlated with systemic inflammation (CRP, IL-6) and symptom severity (1, 17). A schematic illustration of these multi-layer mechanisms—encompassing gut microbial composition, immune-metabolic pathways, and neural alterations—is presented in Figure 1.

Meta-analysis of 10 RCTs (n = 650 adolescents) demonstrated that psychobiotics significantly reduced depressive symptoms vs. placebo (SMD = −0.41; 95% CI: −0.66, −0.16; I² = 49%; p = 0.002; Figure 2B). Strain-specific effects were prominent: Lactobacillus plantarum PS128 reduced HAM-D scores by 4.2 points (Δ = −4.2; 95% CI: −5.1, −3.3; p < 0.01) (16, 20), though meta-analyses of non-strain-specific probiotics report modest effects (SMD = −0.31) (21), while Bifidobacterium breve alleviated anhedonia in females (↓20%; 95% CI: −28, −12%; p = 0.002) (3). Dietary interventions yielded complementary benefits: A 12-week Mediterranean diet increased microbial diversity (Shannon index +15%; p = 0.003) and reduced inflammation (12, 22). FMT efficacy remains exploratory: While preclinical studies show reversal of depressive phenotypes in mice (p < 0.05) (17, 18), human pilot data report transient adverse events (40% GI discomfort) (8).

Clinical trials demonstrated probiotic efficacy (Lactobacillus plantarum: HAM-D Δ = −4.2, p < 0.01), yet safety concerns persist for FMT (40% adverse events). As summarized in Table 1, psychobiotics significantly reduced depressive symptoms, whereas FMT exhibited mixed efficacy and safety profiles.

Publication bias was assessed using Egger’s test (p = 0.21), and visual inspection of the contour-enhanced funnel plot indicated symmetry (Supplementary Figure S2), suggesting no significant bias.

This systematic review makes three pivotal contributions to the literature. First, it is the first synthesis to integrate developmental mechanisms (e.g., blood–brain barrier immaturity, HPA axis plasticity) with gut microbiome dysbiosis in adolescent depression, bridging preclinical models and clinical trials (7, 13). Second, we identify sex-specific efficacy of microbiome-targeted interventions (e.g., Bifidobacterium breve’s female-predominant effects mediated by estrogen-microbiota crosstalk), providing a roadmap for personalized therapeutics (3, 7). Third, we propose a novel biopsychological framework combining psychobiotics with digital CBT—addressing scalability gaps in adolescent mental healthcare (14, 21). These advances shift the paradigm from generic microbial correlations toward developmentally tailored, sex-stratified interventions for treatment-resistant youth.

Our synthesis establishes gut microbiome dysbiosis as a modifiable risk factor in adolescent depression, characterized by inflammation-driven neural dysfunction (hippocampal IL-6↑ 45%, p < 0.01) and neurotransmitter deficits (5-HT↓28%, p = 0.02) (Figure 2) (4, 18). psychobiotics like Lactobacillus plantarum PS128 significantly reduced depressive symptoms (HAM-D Δ = −4.2 vs. placebo, p < 0.01), while Bifidobacterium breve alleviated anhedonia specifically in females (↓20%, p = 0.002) (3, 11). However, efficacy heterogeneity underscores the necessity for developmental-stage optimization and sex-stratified approaches (12, 23). Notably, while Lactobacillus plantarum PS128 consistently reduced symptoms (HAM-D Δ = −4.2; p < 0.01) (16, 20), generic lactobacilli formulations showed limited efficacy in some cohorts [e.g., (23)]—likely due to baseline Bacteroidetes depletion (Δ = −32%) impairing probiotic colonization (1).

The superior response to Bifidobacterium breve in female adolescents may involve estrogen-mediated gut barrier enhancement via ERβ-dependent tight junction upregulation (occludin, claudin-5) (24). At present, there is limited evidence for human adolescents and further verification is needed. Yet, this represents only one facet of sexual dimorphism. Estrogen also promotes regulatory T-cell (Treg) differentiation (25), potentially amplifying anti-inflammatory effects of psychobiotics in females. Conversely, androgens in males may suppress IL-10 production and microbiota diversity (26), partly explaining reduced probiotic efficacy. Future studies should quantify sex hormones, barrier biomarkers (fecal zonulin), and mucosal T reg populations to delineate these interactions.

While FMT from healthy donors reversed depressive phenotypes in adolescent mice (p < 0.05) (17, 19), its human application faces developmental-specific hurdles:

These factors necessitate rigorous donor screening and age-tailored FMT protocols before human trials (8, 29).

Emerging evidence supports the synergistic potential of combining microbiome-targeted therapies with digital mental health platforms for adolescent depression. Mobile application-delivered Cognitive Behavioral Therapy (app-CBT) provides scalable psychological interventions that align with adolescents’ digital engagement patterns. Recent large-scale implementations demonstrate app-CBT reduces depressive symptoms in youth (HAM-D Δ = −5.1, p < 0.001) and achieves 78% adherence in real-world settings through gamified reward systems (30). Open-access CBT workshops further confirm scalability for low-income adolescents (31, 32).

Critically, psychobiotics (e.g., Lactobacillus plantarum PS128) may prime neural circuits for enhanced CBT efficacy by:

This integrated biopsychological approach leverages gut-brain axis modulation to optimize neurocircuitry responsiveness while utilizing digital delivery for scalable skill acquisition—addressing key accessibility barriers in adolescent mental healthcare (14, 22).

The augmentation of CBT efficacy by psychobiotics likely stems from their ability to modulate neurocircuits central to emotion regulation:

Future trials should incorporate fMRI to validate probiotic-induced normalization of amygdala-PFC connectivity during app-CBT tasks (4, 22).

To bridge translational gaps, we prioritize the following: