EFSA opinion on berberine, protoberberines, and plants containing these substances : understanding the draft to engage in the dialogue.
- 2 days ago
- 8 min read
In January 2026, the European Food Safety Authority (EFSA) Panel on Nutrition, Novel Foods, and Food Allergens (NDA) endorsed a draft opinion on the safety of plant preparations containing berberine. This 195-page document, submitted for public consultation, represents a major step in the regulatory oversight of a widely used isoquinoline alkaloid in dietary supplements aimed at metabolic health (blood glucose, lipids, body weight) across Europe.
This article provides a factual and structured reading of the draft, followed by an analysis of areas of uncertainty and levers available to industry stakeholders to actively engage in the regulatory dialogue before the final opinion.
Context and Scope of the Assessment
The request follows the 2019 opinion by ANSES, which identified concerns related to the consumption of plant-based supplements containing berberine: gastrointestinal disorders, hypoglycemia, hypotension, and drug interactions. The European Commission activated the procedure under Article 8(2) of Regulation (EC) No 1925/2006 and asked EFSA to address two key questions:
Is there a link between consumption of the listed preparations and an adverse health effect?
Can a safe daily intake be defined for the general population and vulnerable subgroups?
The mandate covers thirteen species and specific plant parts: Berberis aquifolium, B. aristata, and B. vulgaris (root, bark); Chelidonium majus (aerial parts); Coptis japonica, C. teeta, and C. trifolia (rhizomes); Coscinium fenestratum (root, stem); Hydrastis canadensis (rhizome, root); Jateorhiza palmata (root); Phellodendron amurense (bark); Thalictrum flavum (root); and Tinospora sinensis (root, stem, leaf).
The assessment focuses on the preparations in their entirety, not on berberine alone. Three lines of evidence were integrated: data on isolated berberine, data on other protoberberine alkaloids present in these plants, and data specific to each plant preparation. Explicitly excluded are: benefit-risk analyses, medicinal products, and synthetic forms of berberine (covered under the Novel Food Regulation).
EFSA Conclusions by Toxicological Endpoint
Genotoxicity
Berberine (isolated substance)
EFSA concludes there is compelling evidence of in vitro genotoxicity for berberine. Gene mutations were observed in the HPRT test on murine cells and in the Ames test (S. typhimurium TA98), only without metabolic activation, suggesting a direct mutagenic potential. Clastogenic and/or aneugenic effects were reported in two in vitro mammalian micronucleus tests. Identified mechanisms include DNA intercalation, inhibition of topoisomerases I and II, induction of single- and double-strand breaks, and oxidative DNA damage. Berberrubine, the main phase I metabolite, also inhibits topoisomerase II. In vivo data remain inconclusive: a single mouse study did not confirm these effects, and the Panel emphasizes the need to verify genotoxicity at first-contact sites such as the gastrointestinal tract and liver.
Other Protoberberines
The Panel considers that other protoberberines present in plant preparations may share this genotoxic potential due to strong structural similarity with berberine. QSAR models predict mutagenicity for berberastine, columbamine, epiberberine, jatrorrhizine, palmatine, stephabine, and several others, while experimental data remain sparse and inconclusive, with only isolated signals for coptisine and palmatine.
Chelidonium majus-specific Alkaloids
Sanguinarine and chelerythrine, present in C. majus, pose genotoxic concerns independent of berberine, with evidence of chromosomal and DNA damage in vivo for sanguinarine and QSAR predictions for chelerythrine. These non-protoberberine alkaloids constitute an additional concern.
Carcinogenicity
EFSA establishes evidence of carcinogenicity in rodents for H. canadensis rhizome/root preparations. Two Tier 1 studies show an increased incidence of hepatocellular adenomas in male and female rats, with a positive trend in male mice. Consumption of these preparations therefore represents a carcinogenic risk for humans, even though the exact mechanism remains unclear. A genotoxic role of berberine or its metabolites is possible but not confirmed in vivo. For the other twelve evaluated species, no data are available.
Hepatotoxicity
Berberine (isolated substance)
Hepatotoxicity cannot be established based on available studies. Ninety-day rat studies (156 mg/kg/day) and developmental toxicity studies showed no liver damage, with only two isolated cases of transaminase elevation reported in clinical trials. Berberrubine showed signs of hepatotoxicity in a 42-day rat study at 100 mg/kg/day.
H. canadensis
Tier 1 subchronic studies (90 days) identify the liver as the primary target organ, with rats being the most sensitive species. Increases in liver weight appear at the lowest tested dose (255-260 mg/kg/day), accompanied at higher doses by nearly generalized hepatocellular hypertrophy. These results, consistent with carcinogenicity data, indicate dose-dependent toxicity. EFSA notes that berberine is unlikely responsible, as hepatotoxicity occurs at doses far lower than those used for berberine alone.
C. majus (aerial parts)
Preparations of aerial parts are associated with 43 human cases of idiosyncratic hepatotoxicity, predominantly presenting with jaundice. Latency ranges from a few weeks to several months, making causality difficult to establish. This type of reaction is unpredictable, does not follow a dose-response relationship, and cannot be reliably reproduced in the laboratory.
Developmental and Reproductive Toxicity
Berberine showed signs of maternal and fetal toxicity in rats and mice, with a maternal NOAEL of 223 mg/kg/day in rats and a fetal NOAEL of 666 mg/kg/day in mice. These data are Tier 2 quality. For all plant preparations, information is almost nonexistent, and no reproductive toxicity studies are available, representing a critical data gap.
Systemic Toxicity (Repeated Doses)
No repeated-dose toxicity study compliant with OECD and GLP guidelines is available for berberine alone, preventing the establishment of a regulatory reference point. For most other species, general toxicity profiles are largely unknown. Available studies are Tier 2-3 quality and present limitations such as poorly described test material, a limited number of organs assessed, or incomplete reports.
Drug Interactions
Berberine inhibits CYP3A4 and possibly CYP2D6 and CYP2C9. H. canadensis preparations also show inhibition of CYP3A and CYP2D6 and potential effects on intestinal influx transporters (OCTs), with (−)-β-hydrastine contributing significantly. Preparations containing berberine may therefore interact with many drugs, including anticoagulants, statins, antidiabetics, and antiarrhythmics.
Gastrointestinal Effects
Supplements containing berberine may cause constipation, diarrhea, nausea, or abdominal pain, the most systematically observed signal in clinical trials, at doses of 400-1500 mg/day.
Hypoglycemia, Hypotension, Immunotoxicity
Contrary to previous concerns, no evidence of hypoglycemia, hypotension, or immunotoxicity was found in available animal or human studies.
NDA Panel Conclusion
The Panel concludes: “The available data do not allow for the establishment of a safe intake for any of the plant preparations of the species included in the assessment.”
This conclusion applies to all thirteen species and plant parts of the mandate. It is based on two determinants: established hazard signals (in vitro genotoxicity, H. canadensis carcinogenicity, idiosyncratic hepatotoxicity of C. majus) and a massive insufficiency of data for the majority of species.
Grey Areas and Industry Levers
Although the Panel’s conclusion is strict, a careful reading of the draft reveals an ecosystem of substantial scientific uncertainties, providing real space for dialogue and generation of complementary data. EFSA itself details in Section 6 a structured timeline (Steps 1-4) of information needed to move toward a more differentiated final opinion.
Relevance of in vivo genotoxicity signals
In vitro genotoxicity of berberine is central to the Panel’s concern. However, the draft highlights a major methodological tension: berberine has low systemic bioavailability. Intestinal absorption is limited, intestinal metabolism predominates, and actual systemic exposure is low, with circulating forms mainly as phase II metabolites (glucuro- and sulfoconjugates).
All genotoxicity evidence is based on in vitro systems using free berberine at concentrations that may not reflect actual tissue exposure under real use conditions. EFSA explicitly states that negative in vivo results will only be considered valid if target tissue exposure is demonstrated (toxicokinetic measurements in plasma and tissue homogenates).
Robust tissue toxicokinetic data coupled with well-designed in vivo studies could substantially alter the interpretation of genotoxic risk.
Matrix effect: evaluating preparations, not berberine alone
One of the draft’s key grey areas concerns the matrix effect of plant preparations. EFSA chose to evaluate whole preparations, not berberine alone. While scientifically justified, this creates complexity: the composition of a plant preparation is not limited to its marker alkaloid. Berberine content varies widely depending on botanical origin, plant part, developmental stage, harvest season, extraction process (solvent, temperature, drug/solvent ratio), and analytical method.
Beyond berberine, most co-occurring protoberberines have not been systematically identified and quantified. The unidentified fraction complicates the assessment of mixture genotoxicity. For H. canadensis, the Panel explicitly notes that berberine is likely not responsible for observed hepatotoxicity and carcinogenicity, given the disproportion between equivalent berberine exposure (≈5 mg/kg/day in the hepatotoxic preparation) and doses of berberine alone without effect. Other plant constituents, including (-)-β-hydrastine and canadine, are suggested as potential contributors.
In this context, complete and reproducible chemical characterization of preparations (full alkaloid profile, validated methods, botanical traceability) is the first step required by EFSA (Step 1) and a prerequisite for any regulatory argument.
Non-transposability of data between species
A major difficulty is the lack of data for most species, combined with the temptation to extrapolate from berberine or a better-documented species. The Panel explicitly resists automatic extrapolation, stating that study results for one preparation cannot automatically apply to another, even if berberine content is similar, due to distinct complete alkaloid profiles, differing unknown fractions, and variable component interactions.
This is scientifically coherent but implies that each manufacturer wishing to maintain a species in their portfolio must generate or reference data specific to their preparation. Professional associations (such as EHPM, which has already submitted data in response to the data call) represent an important pooling lever.
EFSA leaves a small door open: “The extent to which results can be extrapolated from one preparation to another will be evaluated based on the data provided and will be subject to expert judgment.” Strong arguments on chemical profile comparability could be presented.
Read-across for protoberberines
Given the absence of experimental data for most protoberberines, the Panel allows for a read-across approach, conditional on applying EFSA SC 2025 guidance. This approach is considered applicable for mutagenicity (given convergent structural alerts and VEGA prediction reliability) but carries high uncertainty for chromosomal effects, due to VEGA model limitations for in vitro and in vivo MN. Experimental data on at least one other protoberberine family member (selected based on a worst-case criterion) are required to reduce uncertainty.
A structured read-across argument supported by PBPK data (physiologically based pharmacokinetic modeling) and in vitro MN data for a worst-case alkaloid could significantly reduce the evidence burden for the entire family.
Public consultation as a dialogue space
With the draft open to public consultation, the window for submitting comments and additional data is a direct lever. Industry stakeholders can:
Submit comments on unclear methodological points (inter-preparation extrapolation criteria, exposure threshold for validating in vivo studies, worst-case definition for protoberberine read-across);
Provide additional analytical data on the alkaloid profile of their preparations;
Contribute to targeted toxicological studies, ideally through consortia via sector associations, for the species best represented in the European market.
Conclusion
EFSA’s draft opinion on berberine and plants containing it is a rigorous scientific assessment but suffers from considerable data asymmetry between species. While the in vitro genotoxic signal and critical cases of H. canadensis and C. majus are legitimate and serious concerns, the inability to establish a safe intake for all thirteen species reflects more a data gap than unequivocal evidence of hazard.
To meet regulatory requirements and support constructive dialogue with authorities, it is essential to provide reliable, well-targeted data, compliant with OECD and GLP standards, and supported by precise chemical characterization of preparations. Within this framework, CEHTRA assists industry by leveraging toxicology expertise to design and conduct rigorous, relevant, and actionable evaluation strategies, enhancing the scientific quality of dossiers and facilitating their assessment.
Author: Marie LIAMIN
References
Draft Scientific Opinion on the safety of plant preparations containing berberine (EFSA-Q-2022-00803). EFSA Panel on Nutrition, Novel Foods and Food Allergens (NDA). 29 January 2026.
