This work is a state-of-the-art narrative review with a targeted and structured literature selection strategy, chosen to address the mechanistic complexity and methodological heterogeneity of gut–thyroid axis research. A formal systematic review was not pursued because of the substantial variability in study design, populations, microbiota assessment methods, and outcome measures, which currently precludes meaningful quantitative synthesis. This approach was chosen due to the complexity of the gut–thyroid axis, which encompasses observational, translational (including animal models), interventional studies, and meta-analyses, alongside substantial methodological and population heterogeneity in the available evidence (8, 21). The aim was not to conduct a formal systematic review, but to provide a critical synthesis of current knowledge from both clinical and mechanistic perspectives. This review was conducted as a narrative, state-of-the-art synthesis; therefore, formal systematic review frameworks such as PRISMA were not applied.

Key publications were identified through searches of PubMed/MEDLINE, Web of Science, and Scopus. The literature search covered publications from January 2009 through December 2025. The search strategy was complemented by manual reference screening (snowballing) of systematic reviews, meta-analyses, and landmark original studies. Combinations of keywords and MeSH terms were used, including “Hashimoto,” “Graves,” “autoimmune thyroid disease,” “gut microbiota/microbiome,” “dysbiosis,” “intestinal permeability,” “short-chain fatty acids,” “probiotics,” “prebiotics,” “synbiotics,” “diet,” “iodine,” “selenium,” and “vitamin D.” Eligible studies addressed gut microbiota composition and function and their metabolites in HT and/or GD, as well as investigations evaluating the effects of dietary and microbiota-targeted interventions on clinical and immunological outcomes. To reduce selection bias, the search strategy prioritized systematic reviews, meta-analyses, and controlled studies where available, and findings were interpreted with explicit consideration of study design, population characteristics, and potential confounders.

Preference was given to publications meeting at least one of the following criteria: (1) studies with a control group, including case–control and cohort designs; (2) analyses examining associations between gut microbiota composition and markers of thyroid autoimmunity (anti-TPO, anti-Tg, TRAb antibodies); (3) randomized interventional trials involving dietary modification, probiotics, prebiotics, synbiotics, or micronutrient supplementation; and (4) systematic reviews and meta-analyses.

Studies of a purely theoretical nature, reports lacking clearly described microbiota analysis methods, and publications in which the effects of AITD could not be disentangled from major confounders—such as recent antibiotic exposure without adequate control—were excluded.

The collected data were organized according to the IMRAD framework. In the Results section, findings are presented by disease entity (HT versus GD), key mechanistic pathways (intestinal barrier function, immunomodulation, microbiota-derived metabolites, molecular mimicry), and type of intervention (dietary approaches, probiotics/prebiotics/synbiotics, supplementation, and other therapeutic strategies). The Discussion provides a critical appraisal of result consistency and study limitations, with particular attention to the impact of dietary variability, geographic factors, and methodological differences across studies (8, 21, 22). Where inconsistencies between studies were identified, emphasis was placed on convergent functional pathways rather than isolated taxonomic findings, in order to minimize overinterpretation of context-dependent results.

In Hashimoto’s thyroiditis (HT), gut microbiota dysbiosis has been consistently reported. The pattern of alterations remains heterogeneous and depends on the studied population, environmental context, and dietary exposures. Meta-analyses and systematic reviews describe disturbances in both alpha diversity and the relative abundance of selected taxa at the genus and species levels, while emphasizing substantial between-study heterogeneity and possible geographic clustering of findings (12, 23).

In a Brazilian cohort, patients with HT showed an increased abundance of Bacteroides and a reduced abundance of Bifidobacterium. Where reported at higher resolution, changes frequently involve genera linked to carbohydrate fermentation and mucosal immune tone (e.g., Bifidobacterium, Lactobacillus, Bacteroides), although the direction of change varies across cohorts. Mechanistically, these shifts are most plausibly interpreted through functional consequences—SCFA output, epithelial integrity, and downstream antigen exposure—rather than a single disease-specific taxonomic signature. Higher circulating zonulin levels were reported alongside microbiota alterations, consistent with barrier perturbation; however, zonulin is an indirect, assay-dependent signal and should be interpreted cautiously (see Limitations of the Evidence) (15, 24–26). Similar observations have been made in other cohorts, although the specific microbial changes differ from one study to another (27). Notably, despite this variability, the reported alterations tend to converge at the level of microbial function rather than individual taxa. This observation reinforces the view that reported taxonomic differences in Hashimoto’s thyroiditis are more likely to reflect underlying functional disturbances of the gut microbiota rather than a single, disease-specific microbial signature.

Against this background, increasing attention has focused on iodine as a potential modulator of both gut microbiota composition and immune responses. Translational studies indicate that excessive iodine intake may disrupt metabolic pathways related to butyrate and propionate metabolism and promote a shift toward a pro-inflammatory immune phenotype that favors thyroid autoimmunity (27, 28). These effects include impaired SCFA synthesis and altered redox balance, providing a functional link between iodine exposure, gut dysbiosis, and immune dysregulation.

Associations have also been reported between specific microbial taxa and the severity of thyroid autoimmunity, as reflected by anti-TPO and anti-Tg antibody titers (12, 23, 28). Interpretation of these relationships is complicated by disease stage, thyroid hormone replacement, and lifestyle modifications. Diet remains one of the main determinants of gut microbiota composition and should be systematically assessed in microbiota studies in HT (15, 23).

In Graves’ disease (GD) and its extrathyroidal manifestation, Graves’ orbitopathy (GO), a marked reduction in gut microbiota diversity is commonly observed. This is typically accompanied by an increased abundance of Bacteroidota and a relative depletion of Firmicutes. These changes correlate with the severity of hyperthyroidism and with higher titers of TSH receptor antibodies (TRAb) (29, 30). In parallel, several studies implicate a pro-inflammatory immune phenotype in active disease, including increased Th17-associated signaling (e.g., IL-17/IL-21) and reduced regulatory tone. This pattern is consistent with enhanced antigen presentation by dendritic cells and macrophage activation in the context of barrier perturbation and altered microbial metabolite profiles.

In patients with Graves’ orbitopathy, several studies have reported gut microbiota changes that tend to track with disease activity and the intensity of the autoimmune response (31–33). Clinically, these findings fit with a broader pattern that includes impaired barrier function, systemic inflammatory features, and progression of orbital involvement (30, 31). From a functional perspective, the available data point toward altered microbial metabolism, which may favor immune profiles associated with disease activity rather than immune regulation.

With effective antithyroid treatment, changes in the gut microbiota appear to move in a more physiological direction, a pattern that tends to coincide with clinical and biochemical improvement. These changes parallel clinical improvement and reduced autoimmune activity (29). In animal models and translational studies, fecal microbiota transplantation from patients with GO attenuated disease manifestations, whereas microbiota modulation, including antibiotics or probiotics, attenuated disease manifestations (34, 35). However, these experimental findings require confirmation in adequately powered human interventional trials.

Probiotic supplementation, particularly with Faecalibacterium prausnitzii, and butyrate administration show therapeutic potential by inhibiting orbital fibroblast activation, reducing fibrosis and adipogenesis, and lowering TRAb levels (36, 37). At the same time, some animal models suggest that specific probiotic strains may exacerbate autoimmunity, underscoring the need for cautious interpretation and further clinical validation (35).

From a clinical perspective, smoking is consistently associated with a more severe course of Graves’ disease and Graves’ orbitopathy, although the underlying mechanisms are likely multifactorial (29, 38). Clinical observations further suggest that managing chronic oral infections may be accompanied by improvement in orbitopathy, pointing to a broader role of mucosal microbial factors in GD/GO (38).

Taken together, dysbiosis in GD and GO is closely linked to disease activity, TRAb titers, clinical course, and treatment response. Environmental modification, including smoking cessation, management of oral infections, and targeted microbiota interventions, may represent valuable adjuncts to conventional therapy (29–33, 35–38).

Taken together, the available evidence suggests that Hashimoto’s thyroiditis and Graves’ disease differ less in specific microbial taxa and more in the way gut microbial function is altered. In Hashimoto’s thyroiditis, these changes tend to cluster around impaired barrier function, disturbances in microbial metabolism, and a background of persistent low-grade inflammation. GD is characterized by reduced microbial diversity, impaired propionate metabolism, and Th17-skewed immune responses. These functional contrasts between HT and GD are summarized in Figure 1.

For clarity, the main differences between Hashimoto’s thyroiditis and Graves’ disease are brought together in Table 1. Rather than listing individual taxa, the table highlights recurring functional features of microbiota alterations that are more relevant from a clinical perspective.

The link between the gut microbiota and thyroid autoimmunity is unlikely to be explained by taxonomic shifts alone. More plausibly, it reflects clinically meaningful functional derangements at the mucosal interface—where environmental exposures are translated into immune dysregulation with downstream endocrine effects. The most consistently implicated mechanisms include altered microbial metabolite output, compromised intestinal barrier function, and immune modulation; acting in concert, these processes may facilitate both the initiation and perpetuation of Hashimoto’s thyroiditis (HT) and Graves’ disease (GD) (Figure 2).

As outlined in Figure 2, diet, micronutrient status, geography, and commonly used pharmacotherapies can reshape key microbial functions, notably short-chain fatty acid (SCFA) production, bile acid metabolism, and epithelial integrity. These functional changes have direct relevance to established immunopathways—shifting the Th17/Treg balance, increasing antigen translocation across a “leaky” barrier, and potentially promoting molecular mimicry—thereby creating conditions that favor the development and persistence of thyroid autoimmunity (6, 10, 11).

Dysbiosis has been linked to AITD through several interconnected mechanistic axes: (1) disruption of the intestinal barrier, (2) immune modulation via the Th17/Treg axis, (3) molecular mimicry between microbial and thyroid antigens, and (4) actions of microbiota-derived metabolites—particularly SCFA and secondary bile acids—acting through G protein–coupled receptors and epigenetic mechanisms (6, 10, 11). Across these axes, a common mechanistic ‘bottleneck’ is mucosal antigen sensing (e.g., LPS–TLR4 signaling), shaping antigen presentation and T-cell polarization (Treg/Th17) with downstream B-cell activation and autoantibody generation.

The thyroid gland and the gastrointestinal tract are closely linked through bidirectional physiological interactions (6, 8, 10). Hyperthyroidism tends to speed up intestinal transit, whereas hypothyroidism slows it. Those motility changes can alter the intraluminal milieu and, in turn, the microbiota profile—one reason cross-sectional associations are difficult to interpret and causality is hard to establish (7, 45). The microbiota may also matter from a pragmatic therapeutic perspective. By shaping enterohepatic handling of hormones and drugs—partly through bacterial β-glucuronidase and sulfatase activity—and by influencing barrier function, bile-acid dynamics, and micronutrient trafficking, it can plausibly affect levothyroxine absorption and overall drug bioavailability (7, 8, 10, 45). In routine practice, it is often the “usual suspects” that have the clearest impact: comorbid conditions such as coeliac disease or Helicobacter pylori infection can impair levothyroxine absorption, and effective treatment has been associated with a clinically meaningful reduction in levothyroxine dose requirement (reported as up to ~34% in some series) (52). Microbiota composition also correlates with the levothyroxine dose needed to maintain stable TSH levels (53).

In AITD, iodine, selenium, iron, zinc, vitamin D, and vitamin B12 are of particular importance for thyroid hormone synthesis and metabolism as well as immune regulation (8, 19, 20, 54–56).

Both iodine deficiency and excess may promote thyroid autoimmunity. Translational data indicate that high iodine intake alters gut microbiota composition and shifts immune responses toward a pro-inflammatory phenotype (6, 54–56). Selenoproteins are critical for thyroid hormone metabolism and antioxidant defense. Selenium supplementation has been linked to lower anti-TPO antibody titers in some cohorts; however, whether this biochemical change results in tangible clinical benefit—such as improved symptoms, reduced levothyroxine requirement, or better long-term thyroid function—remains uncertain (8, 20, 54, 55).

Lower serum 25(OH)D levels are more frequently observed in patients with AITD, but interpretation is limited by multiple confounders (8, 19, 20, 54, 55). Zinc and iron deficiencies, particularly in patients with malabsorption, may impair thyroid function. Iron is a cofactor for thyroid peroxidase (TPO), and iron deficiency can therefore be a clinically relevant contributor to suboptimal thyroid hormone synthesis. In practice, correcting iron deficiency may also improve the response to levothyroxine in selected patients, particularly when persistent symptoms or unexpectedly high dose requirements suggest impaired treatment effectiveness (8, 19, 20, 54–56).

Several recurring methodological and interpretive issues limit the current literature on gut microbiota in AITD. Predominance of cross-sectional studies. Most available data are cross-sectional. They identify associations but do not establish causality (6, 10–12). Prospective studies are scarce. As a result, it remains unclear whether dysbiosis precedes AITD or develops secondary to disease activity and/or treatment (6, 10, 11). MR can partly address temporality using genetic instruments, but MR results should be interpreted cautiously and tested in clinical cohorts (13, 14). As a narrative review, this work is inherently subject to selection bias; however, this limitation was mitigated by a structured search strategy, prioritization of higher-level evidence where available, and a critical appraisal framework emphasizing methodological quality and biological plausibility over isolated positive findings.

Geographic and dietary constraints. Study populations are often geographically homogeneous. This matters because the microbiota is strongly shaped by local diet, environmental exposures, and iodine status (12, 15, 61, 62). For example, microbiota shifts described in a Brazilian HT cohort (increased Bacteroides, reduced Bifidobacterium) may not replicate elsewhere (15). Dietary patterns with established anti-inflammatory and immunomodulatory effects, such as the Mediterranean diet, also modify microbiota composition and may influence AITD risk and disease course (61). Gut microbiota composition and function are strongly shaped by habitual diet and by region-specific exposures, including iodine fortification policies and iodine intake, micronutrient status, and broader environmental factors. Consequently, case–control contrasts observed in one geographic setting may reflect a local baseline microbiota configuration rather than a disease-specific pattern, limiting generalizability across populations. Cross-study comparisons are further confounded by differences in dietary assessment (or lack thereof), variability in fiber and ultra-processed food intake, fermented food consumption, and supplement use, all of which can shift microbial taxa and metabolites relevant to immune regulation. In addition, reverse causation is plausible: patients may change diet after diagnosis (e.g., adopting elimination diets or supplements), which can bias observational associations and inflate apparent “disease-related” microbiota differences. These issues likely contribute to inconsistent directionality of taxa-level findings across cohorts and support emphasizing convergent functional pathways over single taxonomic signatures. Future studies would benefit from standardized dietary assessment, systematic documentation of iodine exposure/status, and multi-center designs with harmonized protocols to improve comparability and external validity.

Geographic and methodological variability. Inter-study heterogeneity in gut microbial findings may partly reflect geographic and methodological variability. Geographic differences in diet, lifestyle, environmental exposures, socioeconomic factors, host genetics, and healthcare practices (including antibiotic and probiotic use) can shape baseline gut microbial communities and may modify observed associations with autoimmune thyroid disease phenotypes. In addition, differences in recruitment criteria and clinical characterization (e.g., disease stage, treatment status, iodine intake, comorbidities) may contribute to inconsistent signals across cohorts. Methodological factors further increase variability, including sample type and handling (stool vs. mucosal samples, storage temperature, freeze–thaw cycles), DNA extraction protocols, primer choice and targeted 16S rRNA regions, sequencing platforms, depth, and bioinformatic pipelines (quality filtering, taxonomic databases, normalization approaches). These sources of batch effects can alter relative abundances and diversity estimates, complicating cross-study comparisons. Future studies would benefit from harmonized protocols, transparent reporting of laboratory and analytic steps, and multi-center designs including diverse populations. Where feasible, standardized pipelines and meta-analytic approaches that account for batch effects and key covariates may improve reproducibility and help distinguish robust disease-related patterns from context-dependent variation.

Incomplete control of confounding. Many studies do not adequately account for BMI, smoking, antibiotic exposure, over-the-counter probiotic use, proton pump inhibitors, or gastrointestinal comorbidities such as celiac disease and IBS, all of which can substantially alter the microbiota (6, 10, 12, 63). Medication effects are another major issue. Commonly used drugs that influence the microbiota (e.g., levothyroxine, metformin, statins, PPIs) complicate interpretation in observational designs (12, 45).

Methodological heterogeneity in microbiota analysis. Differences in analytic approaches—16S rRNA sequencing versus shotgun metagenomics—targeting of different 16S regions, and non-uniform alpha and beta diversity metrics all reduce comparability across studies (6, 12, 64–66). Although 16S rRNA sequencing is widely available and practical for large cohorts, it offers only limited taxonomic granularity and, on its own, says little about what the microbiota is actually doing functionally (64, 66, 67). Shotgun metagenomics provides finer resolution and allows analysis at the pathway level, but the trade-off is higher cost and greater analytic complexity. In practice, shallow shotgun sequencing is often used as a compromise (16, 64–67).

Few interventional studies with clinically meaningful endpoints. Randomized trials with hard clinical outcomes are limited. Examples include changes in levothyroxine requirement, progression of thyroid dysfunction, GD relapse risk, or GO course (10, 11). Most trials focus on surrogate markers such as antibodies or inflammatory indices, often without long-term follow-up (10).

Separating cause from consequence. Distinguishing primary mechanisms from secondary effects is particularly difficult in AITD (6, 7, 10, 68, 69). Hypothyroidism slows intestinal transit, may reduce gastric acid secretion, and can predispose to SIBO. Hyperthyroidism accelerates transit and may cause diarrhea. These effects can reshape the microbiota independent of autoimmunity (68–70). Animal data show that induced thyroid dysfunction alters the microbiota, supporting a bidirectional gut–thyroid relationship (6, 71). Dysbiosis may therefore be part of a feedback loop rather than the initiating trigger (6, 7, 10).

Treatment effects on the microbiota. Levothyroxine and antithyroid drugs may themselves modulate the microbiota, further complicating cross-sectional comparisons (10, 12, 45, 71). Preclinical work suggests some changes may reverse after restoration of euthyroidism, but human clinical data remain limited (10, 71).

Finally, although circulating zonulin has been reported in AITD cohorts and is often discussed in the context of barrier function, its interpretation is limited by assay specificity (several commonly used commercial ELISAs may not reliably quantify pre-haptoglobin 2); therefore, zonulin should be viewed as a non-specific barrier-related signal and ideally interpreted alongside complementary markers (e.g., I-FABP, LPS/LBP/sCD14) or functional permeability testing when available.

Taken together, the data are biologically coherent and clinically interesting, but they are not yet practice-changing. What is needed now are well-designed prospective and interventional studies, with careful control of key confounders and more consistent methodology across cohorts (6, 7, 10, 11).

This review does not argue for microbiota-targeted interventions as disease-modifying therapy in autoimmune thyroid diseases, nor does it propose routine microbiota testing in clinical practice. Instead, it emphasizes careful interpretation of functional signals and their role in supportive, phenotype-informed care.

From a clinical standpoint, the most relevant implications involve interventions with established utility, guided by risk assessment and individualized care (Table 2).

First, clinicians should identify and treat comorbid gastrointestinal disorders, especially celiac disease, which can impair drug and micronutrient absorption. Elimination diets should be implemented primarily when clear indications are present (8, 21, 69). Second, iodine intake and status should be assessed regularly. Both deficiency and excess may exacerbate autoimmunity. Particular caution is warranted in patients using iodine supplements or consuming iodine-rich diets (8, 54).

Next, micronutrient deficiencies should be identified and corrected, including selenium, iron, zinc, and vitamin D. These deficits are common in AITD and may worsen disease expression. Supplementation should be guided by nutritional assessment and laboratory testing rather than routine use (19, 20, 54). Dietary counseling should emphasize a pattern that supports microbiota homeostasis: higher fiber intake (vegetables, fruit, whole grains), moderate use of fermented foods, and reduced intake of ultra-processed foods and saturated fats. A Mediterranean or broadly anti-inflammatory dietary pattern is a reasonable template (8, 19, 72, 73).

Probiotics and synbiotics may be considered in patients with gastrointestinal symptoms or in selected clinical settings. Current data suggest modest benefits, mainly for quality of life and potentially for TRAb reduction in GD, with little effect on thyroid hormone parameters. Routine use in all patients cannot be recommended at present (17, 18, 21). Notably, the same principles that guide pragmatic adjunctive care today should also inform the design of future microbiota-focused studies. For clarity, the key practical implications of the available evidence for clinicians are summarized in Box 1.

Rather than expanding descriptive microbiota catalogues, future progress in this field will depend on more focused and clinically oriented study designs. From a practical perspective, four elements are likely to be critical for generating clinically actionable evidence:

Promising strategies include microbiota- and metabolite-informed personalized nutrition, bacterial consortia, and postbiotics. Their efficacy and safety still require confirmation in clinical trials (6, 10, 62). Future work should move beyond searching for single taxonomic “signatures” and instead stratify patients by microbial function and test interventions against measurable, clinically relevant endpoints.

Shifting the focus from descriptive differences toward functional, phenotype-informed, and outcome-oriented study designs will be essential to translate microbiota research into clinical relevance in autoimmune thyroid diseases.