The Gut-Scalp Axis: A Systematic Review of Dietary Interventions on Microbiome Dysbiosis in Scalp Disorders

The human scalp carries a diverse and functionally significant microbiome that plays a crucial role in the maintenance of skin homeostasis and barrier function. Dysbiosis or alterations in this microbiota, has been associated with a variety of scalp disorders such as dandruff, seborrheic dermatitis, androgenic alopecia (AGA) and alopecia areata (AA) [1-3]. These conditions are typically marked by inflammation, dysfunction of sebaceous glands, follicular disease, and, in some, immune-mediated processes. The scalp microbiota is shaped by intrinsic factors (e.g., genetics, function and immune status) and extrinsic factors (e.g., environmental and shampooing habits), with certain microbial taxa such as Malassezia, Cutibacterium, Staphylococcus and Pseudomonas being differently associated with pathological states [2,4].

Along with these developments, the gut microbiota is increasingly recognized for its role in regulating systemic immunity, metabolic homeostasis and neuroendocrine signaling. There is growing evidence that signals from the gut can modulate skin and hair follicle biology via immunologic and metabolic pathways-referred to as the gut-skin axis [5,6]. Recent extensions of this paradigm have proposed a gut-scalp axis, by which microbial metabolites, cytokine signaling and neuroimmune crosstalk govern scalp-specific inflammatory and regenerative responses [7]. Preclinical and clinical research have demonstrated that oral probiotic therapy and associated microbial interventions can have systemic anti-inflammatory actions, regulate oxidative stress and modulate skin diseases such as atopic dermatitis, psoriasis and acne [6,8].

Emerging data supported a gut-skin-hair axis whereby microbial signals modulated perifollicular immunity, barrier integrity and sebaceous microecology-mechanisms plausibly extending to inflammatory/autoimmune scalp diseases such as Alopecia Areata (AA) and micro‐inflammatory phenotypes that accompany androgenetic alopecia (AGA). In this framework, probiotics, prebiotics, postbiotics, synbiotics and paraprobiotics functioned as microbiome‐directed adjuncts [9-11]. Orally delivered pro/synbiotics were hypothesized to rebalance dysbiosis, elevate short-chain fatty acids and shift T-cell polarization toward Treg dominance while damping Th1/Th17 axes (e.g., IFN-γ/IL-17), thereby reducing perifollicular inflammation relevant to AA; they also potentially mitigated oxidative stress and normalized metabolic mediators that secondarily affect hair cycling [9-10]. Topical paraprobiotics (non-viable microbes) and postbiotics (defined microbial metabolites such as lactic acid, bacteriocins, exopolysaccharides) engaged TLR2/TRL4 and related pattern-recognition pathways in keratinocytes/sebocytes to tighten barrier (↓TEWL, ↑hydration), suppress NF-κB-driven cytokines (e.g., IL-6, IL-31) and favor a scalp biome with lower Malassezia/Cutibacterium overgrowth, without the sensitization or resistance risks seen with chronic corticosteroid or antifungal use [11-13]. Because AGA remains androgen-driven, these modalities were positioned as adjunctive, targeting the micro-inflammatory/oxidative milieu and microbiota imbalance that can exacerbate shedding and symptoms, whereas in AA they may complement immunomodulators by restoring immune tolerance at the follicle. Overall, microbiome-directed strategies were non-invasive, steroid-sparing and mechanistically coherent; however, benefits likely depended on strain/formulation specificity, dose and viability (for probiotics), stability in scalp pH/sebum and host context. Rigorous, adequately powered randomized trials with standardized endpoints (hair density/diameter, SALT or dandruff indices, cytokines, TEWL and taxa/functional profiles) remained necessary to confirm durability, define responder phenotypes and establish where these agents best integrate with existing AA/AGA therapies.

In spite of the rising number of publications, the evidence supporting microbiota-modulating interventions in scalp disease is heterogenous and fragmented. Clinical trials differ by type of intervention, microbial target (scaly scalp vs. gut), diagnostic criteria, follow-up duration and outcome measures. In this context, mechanistic investigation of inflammatory, hormonal or metabolic pathways distal to microbial modulation in scalp disease is not available. Therefore, systematic assessment of the published literature is needed to evaluate if microbiota-targeted dietary and topical interventions result in clinically meaningful, microbial or biomarker-level modifications in scalp health.

The current systematic review was thus conducted to critically assess and synthesize evidence for the efficacy of diet or microbiota-derived approaches to gut or scalp microbiota for scalp disorder treatment, with a focus on clinical effects, microbial modulation, immunomodulation and skin barrier integrity.

Eligibility Criteria

The PECOS (Population, Exposure, Comparator, Outcome, Study design) framework was created to help structure this systematic review in accordance with the PRISMA reporting guidelines [14]. The Population was scalp-related dermatologic condition patients, namely dandruff, seborrheic dermatitis, androgenic alopecia or alopecia areata. The Exposure was microbiome-modyling interventions, including probiotic, prebiotic, paraprobiotic, postbiotic or dietary interventions administered orally or topically. The Comparator was placebo groups, baseline controls or healthy controls, respectively, depending on the study design used. The Outcomes were clinical severity measures (e.g., hair density, sebum secretion and dandruff scales), microbial measures (e.g., alpha/beta diversity and taxonomic changes), inflammatory or hormonal biomarkers, as well as any self-reported symptoms by patients. The Study designs were randomized controlled trials (RCTs), nonrandomized interventional studies and cross-sectional observational studies to allow intervention-based and exploratory microbiome analyses to be eligible.

Inclusion and Exclusion Criteria

Inclusion criteria for studies were those conducted on participants with a diagnosis of a scalp disorder and that examined interventions or correlations between microbiota. Acceptable interventions were formulations, paraprobiotics, postbiotics, prebiotics, probiotics or dietary regimens adjusting microbiota. Studies needed to report at least one outcome for scalp or hair health, clinical, microbial, biochemical or patient-reported. Both observational and interventional study designs were acceptable. Excluded were in vitro studies, animal studies, narrative reviews, editorials, case reports, commentaries and studies in which there was no satisfactory outcome data or modulation of the microbiome was not a primary or secondary outcome. Also excluded were studies not published in full text or in non-English languages.

A range of methodologies were utilized to consider associative and Interventional evidence for the gut-scalp axis. RCTs were selected as they are able to assess the effect of interventions in a controlled setting. Nonrandomized interventional studies were also included to account for realistic, real-life situations where randomization was not feasible or not conducted. Cross-sectional observational studies were considered appropriate to explore possible differences in the microbiome or association with disease compared with controls, particularly where longitudinal intervention data were lacking.

Database Search Protocol

A systematic search was performed in six databases. These consisted of PubMed, Embase, Scopus, Web of Science, Cochrane CENTRAL and ClinicalTrials.gov. Controlled vocabulary (e.g., MeSH and Emtree terms) and free-text relevant terms were incorporated in each database-specific strategy. Boolean operators (AND, OR) were utilized to link search concepts such as scalp disorders (e.g., alopecia, dandruff), microbiota targets (e.g., gut microbiome, skin microbiome) and intervention types (e.g., probiotics, synbiotics). The search strategy was modified to each platform’s syntax to achieve sensitivity and comprehensiveness. No publication date limits were imposed (Table 1).

Table 1: Database-Specific Search Strings

Data Extraction Protocol and Chosen Items

Data were retrieved in duplicate by two reviewers utilizing a pretested, version-controlled template; discrepancies were resolved through consensus or the intervention of a third party. The template encompassed study identifiers (author, year, setting), methodological design and sampling framework, sample size, participant characteristics (age, sex), scalp disorder phenotype, intervention administration (topical, oral, synbiotic; dosage, schedule, duration), target compartment (gut versus scalp microbiota), microbiome assay techniques (16S/shotgun for bacteria, ITS for fungi, qPCR/culture), bioinformatics processes and normalizations (rarefaction/compositional transformations), diversity metrics (α/β), as well as taxonomic and functional outputs. Clinical endpoints (e.g., SALT, hair density/diameter, dandruff indices, sebum), inflammatory and oxidative markers (e.g., IL-6, IL-31, TGF-β1, CRP/hsCRP, SOD) and barrier measurements (TEWL, hydration, pH) were extracted with units standardized a priori (e.g., hairs/cm²; pg/mL; g·m⁻²·h⁻¹). Patient-reported outcomes and adverse events were recorded verbatim and classified into prespecified domains. In cases where multiple time points were available, data closest to the primary window (weeks 4-24) were preferentially abstracted; otherwise, the longest common follow-up period was utilized. Changes from pre- to post-intervention and intergroup comparisons were documented separately; only statistics explicitly articulated in the text, tables or supplementary materials were extracted. Suspected duplicate cohorts were harmonized and ambiguous denominators or derived values were excluded from quantitative synthesis.

Protocol for Assessing Risk of Bias

Risk of bias was appraised with the Joanna Briggs Institute (JBI) tools [15] aligned to design: the 13-item RCT checklist (random sequence generation, allocation concealment, blinding of participants/personnel/outcome assessors, fidelity, complete outcome measurement, appropriate analysis including ITT), the 9-item quasi-experimental tool (baseline comparability, concurrent controls, co-interventions, outcome reliability, follow-up completeness) and the 8-item cross-sectional tool (sampling frame/strategy, adequacy of sample size, confounding identification/control, validity/reliability of exposure and outcome measures). Each item was rated “Yes/No/Unclear/NA” and study-level judgments were derived by domain aggregation, prioritizing internal validity domains (randomization/concealment/ blinding; confounding control; outcome measurement) when discordant.

Evaluation of Evidence Certainty

Certainty of evidence was graded at the outcome level by GRADE [16], initialised at high for randomized trials and low for observational studies, then downward grading for study-level risk of bias (informed by JBI assessments), inconsistency (heterogeneity/non-overlapping CIs), indirectness (mismatch of population/intervention/outcome/ time-point), imprecision (broad CIs across decision thresholds or optimal information size not attained) and publication bias (small-studies effects/asymmetry where possible). Downward upgrading was given consideration for large effects, exposure-response gradients or if likely residual confounding would diminish (rather than inflate) observed effects. Clinical (hair/scalp) outcomes, microbiological endpoints (taxa/diversity), inflammatory/ oxidative markers and barrier measures were graded separately to yield transparent, domain-specification certainty statements.

A systematic search across databases yielded 754 records (Figure 1). After deduplication (28 duplicates removed), 726 unique records were screened. No records were excluded at the screening stage. Of these, 726 full-text reports were sought and 42 were unable to be retrieved. Of the remaining 684 articles, eligibility was ascertained. After assessment, 670 records were excluded, primarily because the records were case reports (n = 186), literature reviews (n = 126), in-vitro studies (n = 164) or failed PECOS criteria (n = 194). 14 studies [17-30] were eventually excluded and added to the systematic review after fulfilling the inclusion criteria.

Figure 1: PRISMA Study Selection Process for the Review

Bias Assessment Observations

The RCTs conveyed a low risk of bias in the majority of areas (Figure 2), including randomization, comparability at baseline, outcome measurement, follow-up and statistical analysis, though some issues related to blinding or confounder management were noted in the majority of trials [17,22,24,26,28]. Woo et al. [29], however, noted a general higher risk because of serious issues in blinding and the occurrence of unresolved confounding variables.

Figure 2: Bias Assessment Across RCTs Included in the Review

Among the cross-sectional studies (Figure 3), Ho et al. [19], Jung et al. [20] and Moreno-Arrones et al. [23] were overall at low risk of bias for outcome measurement and statistical analysis. Ho et al. [19] and Moreno-Arrones et al. [23] also had some issues with confounder identification and adjustment.