Gut microbes and muscle signals may underpin Tuina pain relief, review proposes

A Chinese manual therapy best known as Tuina may owe some of its reported pain-relieving effects to signals traveling far beyond the kneaded tissue — through gut microbes, their metabolites, and hormone-like molecules released by skeletal muscle. That is the central proposal of a narrative review published in Frontiers in Microbiology by researchers at Changchun University of Chinese Medicine and collaborating institutions, who assembled evidence from PubMed and the Web of Science Core Collection up to June 2026. The authors, led by Tuina specialist Jundong Jiao, argue that the brain, gut and skeletal muscle form an interconnected regulatory network — a brain-gut-muscle axis — through which mechanical stimulation applied to skin, fascia and muscle could in principle trigger systemic changes in microbial function, metabolism and neuroimmune signaling. Crucially, however, the review is explicit that this framework remains a hypothesis-generating map rather than an established mechanism: no study to date has continuously linked Tuina stimulation, microbial functional changes, specific metabolites, host receptors and pain improvement within a single experimental design.

Chronic pain is a multilevel phenomenon involving peripheral nociceptive input, central sensitization, neuroimmune responses, metabolic alterations and shifts in gut microecology. Pharmacological options such as nonsteroidal anti-inflammatory drugs, antidepressants, anticonvulsants and opioids vary in efficacy across populations and carry long-term concerns including adverse effects, tolerance and dependence, which has fueled interest in nonpharmacological approaches. Tuina, used mainly for musculoskeletal conditions such as chronic low back pain, neck and shoulder pain, osteoarthritis-related pain and myofascial pain, has shown short-term improvements in pain and function in some randomized trials and systematic reviews of chronic nonspecific low back pain. Yet the review stresses that studies differ markedly in indications, manipulation techniques, stimulation dose, control conditions and methodological quality, and broader massage evidence is largely supported by low- or very-low-certainty findings. Clinical benefit, the authors caution, should be interpreted condition by condition rather than generalized across all chronic pain patients.

The biological core of the proposed framework rests on cross-organ signaling. The central nervous system can influence intestinal motility, barrier function and mucosal immunity through autonomic pathways and the stress axis, while gut microbes and their metabolites can feed back on neuroinflammation and pain processing via circulatory, immune and vagal routes. Skeletal muscle, meanwhile, is both a major site of chronic musculoskeletal pain and an endocrine organ: abnormal mechanical loading, ischemia and mitochondrial dysfunction can heighten nociceptor excitability and sustain peripheral input that drives central sensitization, while contracting muscle releases myokines such as interleukin-6 and irisin, along with lactate and lipid signals, into the circulation. Animal work even suggests muscle-derived signals can reshape the gut — exogenous irisin reduced intestinal inflammation and altered microbial composition in a colitis model — though that study did not test endogenous muscle irisin and did not involve pain.

Where does the gut microbiota fit in? Systematic reviews and meta-analyses have identified alterations in microbial diversity and selected taxa in some chronic pain populations, but the review emphasizes that no stable, broadly generalizable pain-associated microbial signature has emerged. Pain type, geography, diet, medication exposure and analytical methods all muddy the picture. The authors argue that functional relevance may lie less in single genera and more in microbial capacities — short-chain fatty acid production, bile acid transformation, tryptophan metabolism, mucosal barrier maintenance — changing in parallel with pain phenotypes. Some of the strongest functional support comes from outside Tuina research entirely: in neuropathic pain models, butyrate produced by the bacterium Roseburia intestinalis activates GPR41 on vagal neurons, enhancing afferent signaling to the brainstem and modulating a circuit running to the central amygdala, and deleting vagal Gpr41 markedly blunted the analgesic effect. Bile acid receptors TGR5 and FXR in the spinal dorsal horn have likewise been linked to reduced mechanical hypersensitivity after nerve injury, and hyodeoxycholic acid improved neuropathic pain behavior in animals through FXR signaling alongside recovery of intestinal barrier function.

Tryptophan metabolism presents a more tangled picture. Gut microbes can influence three major branches — serotonin, kynurenine and microbiota-derived indoles — and indoles can act as ligands for the aryl hydrocarbon receptor in host-microbial communication. Population studies have associated several tryptophan metabolites with osteoarthritis pain severity, but these findings are predominantly observational and insufficient to establish causality. The review also warns against reading metabolite changes in fixed directions: in a post-inflammatory visceral pain model, shifts in SCFA-producing bacteria were not consistently aligned with actual fecal SCFA concentrations, and the pain phenotype showed marked sex dependence. Concentration, receptor distribution, host immune status and the local microenvironment all modify what a given metabolite does, meaning an increase in one microbial taxon or molecule does not automatically translate into analgesia — or pronociception.

What, then, has actually been observed after Tuina? The direct evidence base is thin and heterogeneous. In mice with experimental ulcerative colitis, Tuina was accompanied by changes in gut microbial composition, colonic protein expression and pathways including biotin metabolism and linoleic-acid metabolism — but the model was an intestinal inflammatory disease, not chronic pain. Abdominal massage in rats with experimental insomnia altered microbial structure and predicted functional pathways alongside brain-gut peptides, again in a non-pain setting. The most pain-relevant human study involved patients with knee osteoarthritis receiving electroacupuncture combined with Tuina, in whom gut microbial changes paralleled improvements in pain and function; because both interventions were given, the changes cannot be attributed to Tuina alone, and the 16S rRNA sequencing used captured community composition rather than direct strain or functional-gene resolution. On the muscle side, massage in rats with skeletal muscle spasm produced 136 differential metabolites tied to mitochondrial and energy-metabolism restoration, while Tuina in a delayed-onset muscle soreness model shifted muscle transcriptomic and metabolomic features over time, with inflammatory mediator regulation of TRP channels, cAMP signaling and ferroptosis among the implicated pathways.

This is where the two omics technologies earn their place in the framework. Metagenomic sequencing directly captures genes within microbial communities, extending analysis beyond 16S-based taxonomic inference to strain-level resolution, functional genes and metabolic pathway potential — including butyrate-synthesis genes such as but and buk, bile salt hydrolase genes (bsh) and bile acid-inducible bai clusters, and tryptophanase (tnaA) driving indole formation. Untargeted metabolomics, by contrast, detects a broad swath of small molecules jointly shaped by host and microbes, making it well suited to discovering previously unrecognized metabolic shifts in mechanistically heterogeneous conditions like chronic pain. The review cautions that functional gene abundance is not equivalent to actual metabolic activity: substrate availability, microbial cross-feeding, intestinal pH and the host environment all shape final metabolite output, and structural identification in untargeted metabolomics remains a major limitation, typically requiring tandem mass spectrometry, authentic standards and targeted quantification for high-confidence results.

Interpreting any post-Tuina omics signal also means wrestling with confounders. Diet and recent food intake, gut transit time, stool consistency, age, sex, BMI, sleep, physical activity and circadian rhythms all shape the microbiome, and patients with chronic pain commonly take NSAIDs, opioids, antidepressants or probiotics that can directly perturb gut ecology. Sampling time itself can alter microbiome comparisons within the same experiment. The review outlines the statistical toolkit needed to cope: MaAsLin2 for covariate-adjusted multivariable modeling, ANCOM-BC and ANCOM-BC2 for compositional differential abundance, mixed-effects models for repeated measures, and false discovery rate control across high-dimensional features. On the intervention side, sensor-based studies show that Tuina parameters such as force waveform, peak force and frequency vary measurably between practitioners — in Ding rolling manipulation, for instance — so the authors call for quantitative recording of mechanical exposure alongside reporting frameworks like STRICTO for Tuina trials and STORMS for microbiome studies.

The bottom line is a research agenda rather than a verdict. Because multi-omics associations reflect synchronous variation and remain vulnerable to sample size, confounding, batch effects and collinearity, the authors argue that confidence can only build through longitudinal multi-time-point sampling, targeted metabolite quantification, replication in independent cohorts, and functional perturbation — microbiota transplantation, specific strain supplementation, metabolite intervention and receptor blockade of the kind that validated the butyrate-GPR41 pathway. No prospectively validated microbiome or metabolite biomarker panel currently predicts response to Tuina, and machine-learning models built on high-dimensional omics data can lose substantial performance when applied across cohorts. Lifestyle factors such as diet, exercise and sleep already have broader roles in comprehensive chronic pain care, whereas microecological interventions targeting specific taxa or metabolites remain strongly exploratory. The brain-gut-muscle axis, the review concludes, offers a testable cross-organ scaffold for investigating the systemic effects of Tuina analgesia — but its specific pathways and clinical predictive value still await direct experimental and prospective validation.

Subject of Research: A narrative review proposing a brain-gut-muscle microecological-metabolic framework for investigating Tuina analgesia in chronic pain

Article Title: Microecological-metabolic links along the brain-gut-muscle axis in Tuina analgesia from the perspectives of gut metagenomics and untargeted metabolomics

Article References: Zhang, X., Wang, H., Xing, J., Zhao, H., Liu, Y., & Jiao, J. (2026). Microecological-metabolic links along the brain-gut-muscle axis in Tuina analgesia from the perspectives of gut metagenomics and untargeted metabolomics. Frontiers in Microbiology, 17, Article 1957609. https://doi.org/10.3389/fmicb.2026.1957609

Image Credits: AI Generated

DOI: 10.3389/fmicb.2026.1957609

Keywords: Tuina, chronic pain, brain-gut-muscle axis, gut microbiota, metagenomics, untargeted metabolomics, short-chain fatty acids, bile acids, tryptophan metabolites, myokines, central sensitization, manual therapy