Ancient Herb Meets Acupuncture: Scientists Map Shared Liver-Defense Pathways — but Urge Caution on Combining Them

Liver disease kills roughly two million people each year, about four percent of all deaths worldwide, and the therapies that exist today mostly target the cause of injury rather than the scarring it leaves behind. Once fibrosis takes hold, approved drugs that reliably reverse it remain scarce and specific to individual indications. Against this backdrop, a team of Chinese researchers has published an unusually candid synthesis asking a question that traditional and modern medicine have rarely confronted side by side: do a classic herbal remedy and a classical needling technique actually protect the liver through overlapping biology? The review, published in Frontiers in Molecular Biosciences, examines saikosaponins, the triterpenoid saponins of Radix Bupleuri, alongside acupuncture, electroacupuncture, moxibustion and acupoint catgut embedding. Its conclusion is nuanced and deliberately restrained: the two intervention streams show partial functional overlap in several disease-relevant domains, but combining them remains a falsifiable hypothesis, not a demonstrated therapy.

The choice of herb is not arbitrary. Bupleurum chinense and Bupleurum scorzonerifolium have anchored East Asian liver-directed formulas such as Xiao-Chai-Hu-Tang and Chaihu-Shugan-San for centuries, and their best-characterised constituents, the saikosaponins, carry one of the larger preclinical hepatology literatures in natural product research. The most studied congeners are saikosaponin a and saikosaponin d, with saikosaponins b1 and b2 following behind. Across chemically induced, immune-mediated, alcohol-related, cholestatic and metabolic models of liver injury, these molecules have been shown to suppress lipid peroxidation, lower transaminases, dampen inflammatory signalling and restrain the activation of hepatic stellate cells, the myofibroblast-like cells widely regarded as the central drivers of fibrogenesis. In carbon tetrachloride injury models, for instance, saikosaponin d and saikosaponin a reduced oxidative stress and NLRP3 inflammasome activation, while in fibrotic models saikosaponin d cut collagen I deposition and transforming growth factor beta 1 expression.

Yet the review is emphatic that this breadth of models should not be mistaken for uniform evidence. Saikosaponin pharmacology is congener-specific, meaning that structurally near-identical molecules act through partly distinct molecular nodes. Saikosaponin d works largely through an estrogen receptor beta axis and the ROS/NLRP3 inflammasome to deactivate stellate cells, whereas the ring-opened congener saikosaponin b1 attenuates fibrosis by disrupting the STAT3/Gli1 interaction and promoting Gli1 degradation, and does so with lower hepatocyte toxicity. Saikosaponin a favours LXR-alpha-dependent and TLR4-directed anti-inflammatory routes, blocking the translocation of TLR4 into membrane lipid rafts by depleting membrane cholesterol. Gut microbial and hepatic deglycosylation further complicates the picture, because metabolites such as saikogenin A, which relieves alcoholic steatosis through SIRT1 signalling, may differ from the administered parent compounds in potency, target profile and cytochrome interactions. The active species in the body may simply not be the molecule in the bottle.

The safety picture is the decisive caveat for translation. Saikosaponins display a well-recognised dual, dose-dependent nature: at sub-toxic doses, saikosaponin d aggravated acetaminophen-induced liver injury by binding GABARAP and obstructing SNARE-dependent autophagosome maturation, sabotaging the very protective autophagy that mediates benefit in other contexts. Saikosaponins also worsened cholangiocyte mitochondrial damage in cholestasis by inhibiting protective autophagy. On top of this, saikosaponins and their metabolites modulate CYP3A4 through the pregnane X receptor, and saikosaponin d alters CYP1A2, CYP2D6 and CYP3A4 in HepaRG cells, creating a realistic drug-interaction risk given that CYP3A4 metabolises a large share of clinical medicines. Poor aqueous solubility and low oral bioavailability compound the problem, and surfactant-like membrane activity limits parenteral use through haemolysis. Delivery engineering, including liposomal formulations that reduced haemolysis and widened the gap between effective and toxic doses in fibrotic mice, currently offers the most credible route to a favourable benefit-risk balance, though almost all pharmacokinetic data remain rodent-based.

The acupuncture side of the analysis is equally disciplined about distinguishing modalities. Manual needling, electroacupuncture, moxibustion and acupoint catgut embedding differ in stimulus, dose and plausible physiology, and the evidence is distributed unevenly among them. Human evidence is strongest for metabolic liver disease and concentrated in electroacupuncture and embedding protocols. A randomised controlled pilot trial reported that electroacupuncture reduced hepatic fat measured by MRI-PDFF in non-alcoholic steatohepatitis, outperforming sham stimulation, and a further randomised trial found electroacupuncture added to lifestyle modification improved steatosis and glucolipid metabolism in obese NAFLD patients. Meta-analyses of thirty randomised trials in NAFLD and of acupoint embedding studies reported improved response rates, transaminases and lipid profiles. For advanced disease the picture weakens sharply: a cirrhosis meta-analysis judged the evidence quality low, and a Cochrane review of acupuncture in chronic hepatitis B rated certainty very low.

Mechanistically, the most striking recent work has given acupoint selection a testable neuroanatomical foundation. Researchers showed that stimulation at the classical point Zusanli, ST36, but not at abdominal points, recruits PROKR2-marked sensory neurons that drive a vagal-adrenal anti-inflammatory circuit, and only above a threshold of afferent activation. Building on this, electroacupuncture pretreatment at ST36 protected mice against hepatic ischemia-reperfusion injury through a vagus-dependent activation of the Nrf2 antioxidant pathway, a finding causally confirmed by vagotomy experiments. Electroacupuncture has also been linked to NRF2/ARE activation, AMPK-dependent glycolipid metabolism and autophagy regulation in metabolic liver models, while ST36 moxibustion has been tied to AMPK/mTOR-regulated hepatic autophagy. The authors caution, however, that cholinergic anti-inflammatory signalling has been demonstrated most rigorously in intestinal rather than hepatic macrophages, and that reductions in whole-liver alpha-SMA and collagen I are consistent with lower fibrogenic activity without proving direct stellate cell deactivation.

Placed side by side, the two evidence streams converge on four functional domains rather than a single shared molecular network. Both appear to influence fibrogenic cell state and matrix turnover, oxidative-stress control, innate-inflammatory signalling and metabolic-autophagic regulation. Critically, the authors describe this convergence as asymmetric. Vagus-dependent hepatic Nrf2 activation is causally supported for electroacupuncture, whereas direct evidence that saikosaponins activate canonical hepatic Nrf2/HO-1 signalling remains limited, so Nrf2 cannot be claimed as an equivalently established shared node. Likewise, AMPK is not established as a common saikosaponin target, and the autophagy story is directionally treacherous, since the same molecule can restrain autophagy in fibrotic stellate cells while promoting autophagosome formation to trigger their apoptosis, and can block protective autophagy in injured hepatocytes. A fifth, cross-cutting interface, the gut-liver axis, links the streams still more loosely: saikosaponin d has altered gut microbiota and bile-acid metabolism in NAFLD models, while electroacupuncture affects microbiota, intestinal barrier integrity and gut-brain signalling, but few studies establish causal mediation.

What emerges is less a treatment claim than an experimental roadmap. Because no direct co-administration study was identified anywhere in the literature, the authors insist the combination be tested in a prespecified two-by-two factorial design: vehicle or sham, a defined purified saikosaponin formulation, a fully reported and reproducible acupuncture protocol, and the combination, ideally in a clinically relevant metabolic-fibrosis model. Interaction must be estimated formally rather than inferred from pathway overlap, because the joint response could just as plausibly be antagonistic or redundant as additive, particularly where both interventions push saturable inflammatory outputs or steer autophagy in opposite directions in different cell types. Co-exposure studies must track plasma and hepatic drug exposure, liver injury markers, autophagic flux and cytochrome substrate interactions, and acupuncture parameters should be fixed and reported under STRICTA guidelines so that an under-specified protocol cannot masquerade as an absence of effect.

The review’s final verdict is a model of scientific restraint in a field prone to hype. Selected saikosaponins and defined acupuncture modalities show genuinely overlapping downstream effects in fibrotic, redox, inflammatory and metabolic biology, with the gut-liver axis as a candidate modifier awaiting causal testing. But the evidence is structurally lopsided: the herbal data are overwhelmingly preclinical and constrained by congener specificity, model limitations, poor bioavailability and toxicity, while the acupuncture data are modality-dependent, with human proof strongest for metabolic surrogates rather than histological fibrosis reversal. Computational tools such as network pharmacology and multi-omics can nominate candidate interactions but cannot validate them. Until factorial experiments with pharmacokinetics, safety assessment and mechanistic perturbation are run, and early-phase human studies follow, combining Chaihu’s molecules with the needle remains exactly what the authors call it: a rigorous, testable hypothesis about how two ancient interventions might converge on the modern diseased liver.

Subject of Research: Shared hepatoprotective mechanisms of saikosaponins from Radix Bupleuri and acupuncture-related interventions and the rationale for combined liver therapy

Article Title: Bridging herb and needle: shared hepatoprotective mechanisms of saikosaponins and acupuncture and the rationale for combined therapy

Article References: Xu, Y., Tang, L., Lai, M., Yang, Y., Zhang, L., & Tian, X. (2026). Bridging herb and needle: shared hepatoprotective mechanisms of saikosaponins and acupuncture and the rationale for combined therapy. Frontiers in Molecular Biosciences, 13, Article 1949938. https://doi.org/10.3389/fmolb.2026.1949938

Image Credits: AI Generated

DOI: 10.3389/fmolb.2026.1949938

Keywords: saikosaponins, Radix Bupleuri, acupuncture, electroacupuncture, liver fibrosis, hepatoprotection, NAFLD, hepatic stellate cells, Nrf2, autophagy, gut-liver axis, traditional Chinese medicine