Pyrrolizidine Alkaloids in Butterbur: Why the Standard EU Panel Misses Petasites japonicus
The short answer
Butterbur contains pyrrolizidine alkaloids (PAs), compounds that can damage the liver and are treated as genotoxic carcinogens. In Japanese butterbur, Petasites japonicus, the main ones are petasitenine, neopetasitenine and senkirkine. They are water-soluble and can be removed during processing, but only testing each finished lot, for the right alkaloids, shows that the removal worked.
A certificate of analysis for a butterbur extract can say two true things at once. The lot complies with the EU maximum level for pyrrolizidine alkaloids. And two of the three alkaloids this plant makes in the largest amounts were never measured.
Both statements can hold because of how the EU list was assembled. It was drawn up mainly to catch contamination, PA-producing weeds harvested alongside tea, herbs and spices, from the alkaloids that food monitoring was finding in Europe. Butterbur is not contaminated with pyrrolizidine alkaloids. It makes them, and most of what Petasites japonicus makes is not on the list.
For this species, a PA result is only as meaningful as three things behind it: which alkaloids were measured, how low the laboratory could see, and whether the finished lot was tested.
This guide to butterbur pyrrolizidine alkaloids covers what the EU list was built to catch, which of P. japonicus's own alkaloids it can and cannot see, why "not detected" can mean anything from zero to a sizeable upper bound, where in the process testing belongs, what "PA-free" means across the EU, US and Japan, and how to read a butterbur PA certificate line by line.
01 Framing
What the EU List Was Built to Catch
The EU maximum levels for pyrrolizidine alkaloids were first set by Regulation (EU) 2020/2040 and now sit in Regulation (EU) 2023/915. The maximum level applies to a lower-bound sum of 21 named alkaloids, plus 14 isomers that can co-elute with them in current analytical methods.[1] Food supplements containing herbal ingredients, extracts included, carry a maximum of 400 µg/kg.[1]
Most of that framework is aimed at contamination. The regulation's recitals say that PAs in these foods can be minimised or prevented through good agricultural and harvest practices, and that maximum levels exist to make sure those practices are applied in every production region.[1] Tea, chamomile and cumin do not make pyrrolizidine alkaloids. They pick them up from ragwort, heliotrope and borage-family weeds that are cut and dried with the crop.
The regulation does not ignore plants that make their own PAs. Borage has limits of its own, 750 µg/kg for fresh leaves and 1,000 µg/kg for dried borage, lovage, marjoram and oregano.[1] What decided which alkaloids went on the list, though, was what European monitoring was finding and which analytical standards existed: the Commission's original selection of 28 alkaloids was built on EFSA's 2011 opinion, two food monitoring projects and the reference standards then available.[2] The alkaloids of P. japonicus were not part of that picture.
Butterbur therefore falls outside the framework twice. Its alkaloids are intrinsic, so no harvest practice removes them; only processing can. And its main alkaloids are not the ones the list was assembled around. The 400 µg/kg limit still applies to a butterbur extract sold into an EU food supplement, but meeting it tells you about the listed compounds only.
02 Compounds
What Japanese Butterbur Makes, and What the List Sees
Three alkaloids, petasitenine (also called fukinotoxin), neopetasitenine and senkirkine, account for roughly 99% of the pyrrolizidine alkaloids in P. japonicus.[3] Neopetasitenine is present in fuki at high enough concentration that a 2021 toxicokinetic study chose it as its model compound for that reason.[4]
One caution about the literature: almost all published PA data for this species come from the edible aerial parts, the petioles, young flower spikes and buds. The rhizome, the raw material for root extracts, is far less documented. That is one more reason the finished extract, not the literature, has to be tested.
| Alkaloid | Necine Type | On the EU 21+14 List | Forms an N-Oxide | Why It Matters |
|---|---|---|---|---|
| Petasitenine (fukinotoxin) | Otonecine | No | No | Carries the plant's carcinogenicity record: liver tumours in rats given 0.01% in drinking water[5] |
| Neopetasitenine (acetylfukinotoxin) | Otonecine | No | No | Present at high concentration in fuki, and rapidly converted to petasitenine after oral intake in rats[4] |
| Senkirkine | Otonecine | Yes | No | The only otonecine-type alkaloid on the list |
The list is not blind to this species. It sees senkirkine, and the way senkirkine appears in the regulation is itself a clue. Almost every other entry is listed as a pair, the alkaloid and its N-oxide, while senkirkine is listed alone.[1] Otonecine-type alkaloids carry a methylated nitrogen that cannot be N-oxidised, so there is no partner to list. The regulation already accommodates one alkaloid of this class. It simply stops there.
A 21+14 result on Petasites japonicus is a measurement of one of the species' three main alkaloids.
Measuring petasitenine without neopetasitenine would not be a conservative shortcut either. After oral dosing in rats, neopetasitenine was absorbed quickly and converted to petasitenine, which then cleared slowly from plasma.[4] Petasitenine is the compound with the carcinogenicity record: in the 1977 study, eight of ten rats that survived beyond 160 days on a 0.01% solution developed liver tumours.[5] Neopetasitenine is best understood as petasitenine in waiting.
Adding the two is a method decision, not a technical barrier. Petasitenine is a catalogued compound (CAS 60102-37-6), and a 2022 Japanese study quantified all three alkaloids in cultivated fuki by LC-MS/MS.[3] A supplier that tests only the EU panel on this species has chosen not to extend it.
03 Arithmetic
Lower Bound, Upper Bound and the Number on the Certificate
The EU sum is calculated on a lower-bound basis: any alkaloid below the limit of quantification counts as zero.[1] Two variables therefore decide the number on the certificate, the analyte list and the LOQ, and neither is visible in the number itself.
How far apart laboratories can be is documented. In the occurrence data submitted to EFSA for tea and herbal infusions, reported LOQs ranged from 0.5 µg/kg for heliotrine to 151.8 µg/kg for retrorsine, and some data sets reported high LOQs for every alkaloid they analysed.[2] The example below uses invented figures to show what that spread does to two honest reports on the same material. It describes no real lot.
| Illustrative Extract: Petasitenine 30, Neopetasitenine 20, Senkirkine 3 µg/kg | Laboratory A | Laboratory B |
|---|---|---|
| Analytes | EU 21+14 | EU 21+14 plus petasitenine and neopetasitenine |
| LOQ per analyte | 5 µg/kg | 1 µg/kg |
| Senkirkine | Below LOQ, counted as 0 | 3 µg/kg |
| Petasitenine | Not measured | 30 µg/kg |
| Neopetasitenine | Not measured | 20 µg/kg |
| Reported result | 0 µg/kg, "not detected" | EU sum 3 µg/kg; extended total 53 µg/kg |
Both reports are true, and both lots comply with the EU limit. Only Laboratory B's report describes the material.
Laboratory B's result also raises the question of how the extended analytes should be reported. Folded into a single total, they can only push that total up, so the figure is conservative for EU compliance: the regulated sum can never be higher than a total that contains it. What a single total cannot show is which alkaloids drove it. The useful question to put to a supplier is therefore not only whether the number is low, but whether it includes petasitenine and neopetasitenine.
What "nothing quantified" can mean
The lower bound is the rule for compliance, but it is not the only way to read a result. For exposure assessment, EFSA works with both bounds: at the lower bound, results below the LOQ are set to zero; at the upper bound, they are set to the LOQ itself.[2] For a lot in which nothing was quantified, the lower-bound sum is always zero, and the upper-bound sum is the LOQs added together. That second number depends entirely on how sensitive the method was.
At an LOQ of 10 µg/kg, a lot in which nothing was quantified could, at the upper bound, hold more than half the EU limit for supplements. At 0.5 µg/kg, the same "not detected" rules that out. When a total is reported as "< X", it is worth asking whether X is the sum of the individual limits. If it is, the result says that nothing was quantified, and X is a ceiling rather than a measurement.
Certificates also use three phrasings that are easy to read as one. "Not detected" means no signal above the limit of detection. "Below LOQ" means there may be a signal, but it is too small to put a number on. A bare figure means the alkaloid was quantified. Each phrasing is only useful alongside the value it refers to.
04 Testing point
Raw Material or Finished Lot
Extraction concentrates. A clean result on dried rhizome says little about the extract made from it, because a kilogram of extract carries the alkaloid load of several kilograms of plant, minus whatever the process removed.
The EU rules recognise the principle for tea: powdered tea extracts are assessed with a concentration factor of four.[1] Food supplements are handled differently. The limit applies to the product as sold, so the formulator's inclusion rate is part of the compliance calculation.
At a release specification of 35 µg/kg, 50 mg of root extract in a 500 mg capsule contributes at most 3.5 µg/kg to the finished supplement, under 1% of the EU maximum. A leaf formula using 133 mg of extract in the same capsule reaches at most 9.3 µg/kg, about 2%. Batch results sit below the specification, so these figures are ceilings, not estimates.
Removal is possible because of a property of the alkaloids themselves: the PAs in P. japonicus are water-soluble.[3] That is why the traditional Japanese practice of boiling and soaking fuki lowers them; the data are covered in Petasites japonicus vs Petasites hybridus: Japanese and European butterbur compared. At production scale, an acid-water wash exploits the same property. The alkaloids protonate and move into the aqueous phase, while the lipophilic sesquiterpenes that carry the product's value stay behind. Removal is not the same as verification, and verification belongs on the finished lot, not on the raw material or an intermediate.
Plant part matters as much as process. Even within the edible aerial parts, the Japanese data show young flower spikes carrying several times the alkaloid level of petioles, and cultivation method alone producing roughly a threefold difference between petiole samples.[3] Root and leaf lines should therefore be run and tested as separate products, never under one PA assumption. How the root line is specified is covered in Total Petasins explained: petasin, isopetasin and how to specify butterbur extract, and the leaf line in Bakkenolide B and butterbur leaf extract.
05 Markets
What "PA-Free" Means in Each Market
No regulation defines "PA-free". Each market has a limit, a method or a presumption instead, and they do not line up.
EU food law sets a concentration: 400 µg/kg for herbal food supplements, on the 21+14 lower-bound sum.[1]
EU medicines work on daily intake instead. A herbal medicinal product must keep adult intake at or below 1.0 µg of PAs per day, and testing is expected to cover a longer list of 28 alkaloids.[6] Those 28 are the same set the Commission first selected for food monitoring.[2] The list includes senkirkine and, like the food list, neither petasitenine nor neopetasitenine. Moving a butterbur ingredient from the supplement route to the medicines route changes the limit structure, but it does not close the gap in what is measured.
The daily-intake view is a useful cross-check even for supplements. At 35 µg/kg, a 50 mg daily serving of extract delivers at most 1.75 ng of PAs, less than 0.2% of the medicinal limit.
Japan, where fuki is an everyday vegetable, frames the question through preparation rather than a number. The agriculture ministry sees no known health concern from eating it, on condition that it is prepared in the traditional way and not overeaten.[3] That position covers a boiled and soaked vegetable at dietary intake. It says nothing about an unprocessed extract.
The United States treats a dietary supplement that contains PAs as adulterated. The FDA has indicated that ingredients from PA-containing plants can still be used if they have been processed and confirmed PA-free by validated analytical methods. No validated compendial method exists yet, and USP has one under consideration.[7] The American Herbal Products Association goes further for its members. Products with botanical ingredients that contain toxic PAs, with Petasites spp. named on the list, may not be sold for internal use and must carry an external-use-only statement.[8]
For a US brand making an oral butterbur product, the practical route is to show that the ingredient no longer contains toxic PAs. That depends entirely on the analyte list, the LOQ and the testing point. The phrase on the label proves none of it.
"PA-free" is a claim, not a result. The question a buyer can answer is: below what, across which alkaloids, measured where.
Removing the alkaloids does not close every safety question about butterbur. The UK's 2012 warning concerned extracts from which PAs had largely been removed, which points to the extract's sesquiterpenes rather than to its alkaloids. That is a question of petasin dose, separate from PA testing.
06 Certificate
Reading a Butterbur PA Certificate
Five things decide whether a PA line on a butterbur certificate means anything.
The analyte list, by name. "Pyrrolizidine alkaloids and PA N-oxides" describes a category, not a scope. For P. japonicus, look for petasitenine and neopetasitenine named as analytes. If they are absent, the result covers senkirkine and the weed panel only.
The LOQ, per analyte, and the summing basis. A lower-bound sum at an LOQ of 1 µg/kg and one at 10 µg/kg can both read "not detected" and describe very different materials.
The testing point. A finished-lot result describes what ships. A raw-material or intermediate result describes something else.
Whether the total includes the extended analytes. A total that includes petasitenine and neopetasitenine is a conservative figure for EU compliance, because it can only overstate the regulated sum. A total that leaves them out says nothing about the species' own main alkaloids. The certificate, or the supplier, should say which it is.
Who tested, and to what standard. Say whether the laboratory is in-house or an accredited third party, and whether the method follows the approach of Ph. Eur. 2.8.26.
PhytoVale's two most recent root-line batches, on those terms:
| FD-2026081209 | FD-2026091501 | |
|---|---|---|
| Plant part | Rhizome and root | Rhizome and root |
| Manufactured | August 2026 | September 2026 |
| Analytes | EU 21+14 plus petasitenine and neopetasitenine | EU 21+14 plus petasitenine and neopetasitenine |
| Reported as | One total, both extended analytes included | One total, both extended analytes included |
| Method | LC-MS/MS | LC-MS/MS |
| Specification | ≤ 2 µg/kg | ≤ 35 µg/kg (≤ 0.035 ppm) |
| Result | < 2 µg/kg | < 28 µg/kg (< 0.028 ppm) |
| Testing point | Finished lot | Finished lot |
| Laboratory | In-house QC | In-house QC |
Because both totals include petasitenine and neopetasitenine, each result is also an upper bound on the EU-regulated sum for that lot.
PhytoVale grows Petasites japonicus on its own cultivation base in Henan: approximately 33 hectares, in production since 2018. Root and leaf are separated at harvest and run as separate lines, each with its own PA results. Grades, specifications and batch data are on the Petasites japonicus extract product page, and every sample ships with its batch certificate. Samples are free, MOQ is 1 kg, and quotations are returned within 24 hours.
07 Questions
Butterbur Pyrrolizidine Alkaloids: Frequently Asked Questions
Is butterbur banned in the UK or the EU?
Not as a plant. The EU regulates the PA content of food supplements rather than banning butterbur, and the regulation explicitly leaves room for stricter national rules on placing PA-containing plants on the market, so status can differ between member states. In the UK, the MHRA advised consumers in 2012 not to take unlicensed butterbur products after reports of liver toxicity; licensed herbal medicines follow a separate route. Check the destination market before formulating.
How much pyrrolizidine alkaloid is considered safe?
Regulators do not set a safe intake for the cancer risk, because 1,2-unsaturated PAs are treated as genotoxic carcinogens. EFSA instead calculates a margin of exposure against a reference point of 237 µg/kg body weight per day, and a margin of 10,000 or more is regarded as low concern. For shorter-term liver effects, Germany's BfR uses a health-based guidance value of 0.1 µg/kg body weight. Legal limits are set partly on what good practice can achieve, not on toxicology alone.
What does "PA and PA N-oxides" mean on a certificate of analysis?
Many PAs occur in plants partly as N-oxides, an oxidised form. N-oxides are only absorbed after gut bacteria reduce them back to the parent alkaloid, which is why regulators count both forms. On a certificate, the phrase names a class of compounds. Which ones were actually measured, and to what limit, has to be stated separately.
Do drying or heating destroy pyrrolizidine alkaloids?
Not reliably. PAs survive drying, which is why the EU sets limits for dried teas and herbs. In the Japanese fuki work, boiling petioles for three minutes and then peeling them still left 66–81% of the alkaloids. The large reductions came from soaking afterwards, as the alkaloids dissolved out into the water. Removal depends on separation, not heat.
Do pyrrolizidine alkaloids build up in the body?
The concern is cumulative damage rather than storage. Repeated exposure can cause hepatic veno-occlusive disease, which appears after weeks to months, while acute liver failure needs doses in the mg/kg body-weight range. Germany's BfR treats exposure beyond 14 days as chronic, which is why regulators assess daily intake over time rather than single servings.
Which other herbs contain pyrrolizidine alkaloids?
Comfrey, borage, coltsfoot, lungwort, European ragwort, groundsel and heliotrope are among the plants named alongside butterbur in the AHPA trade requirement, together with Crotalaria, Eupatorium and several borage-family genera. The same restriction on internal use applies to all of them unless the PAs have been removed and the removal verified.
08 Evidence base
References
[1] Commission Regulation (EU) 2020/2040 of 11 December 2020 amending Regulation (EC) No 1881/2006 as regards maximum levels of pyrrolizidine alkaloids in certain foodstuffs. Provisions now consolidated in Regulation (EU) 2023/915. EUR-Lex
[2] European Food Safety Authority (2016). Dietary exposure assessment to pyrrolizidine alkaloids in the European population. EFSA Journal, 14(8), 4572. DOI: 10.2903/j.efsa.2016.4572
[3] Takenaka, M., Miyake, N., Kimura, T., Todoriki, S., Urushiyama, T. (2022). Reduction of pyrrolizidine alkaloids by cooking pre-treatment for the petioles and the young spikes of Petasites japonicus. Food Science and Technology Research, 28(3), 245–255. DOI: 10.3136/fstr.FSTR-D-21-00245
[4] Yanagi, M., Kamiya, Y., Murayama, N., Banju, K., Shimizu, M., Yamazaki, H. (2021). Metabolic profiles for the pyrrolizidine alkaloid neopetasitenine and its metabolite petasitenine in humans extrapolated from rat in vivo and in vitro data sets using a simplified physiologically based pharmacokinetic model. Journal of Toxicological Sciences, 46(9), 391–399. DOI: 10.2131/jts.46.391
[5] Hirono, I., Mori, H., Yamada, K., Hirata, Y., Haga, M., Tatematsu, H., Kanie, S. (1977). Carcinogenic activity of petasitenine, a new pyrrolizidine alkaloid isolated from Petasites japonicus Maxim. Journal of the National Cancer Institute, 58(4), 1155–1157. DOI: 10.1093/jnci/58.4.1155
[6] European Medicines Agency, Committee on Herbal Medicinal Products. Public statement on the use of herbal medicinal products containing toxic, unsaturated pyrrolizidine alkaloids (PAs) including recommendations regarding contamination of herbal medicinal products with PAs. EMA/HMPC/893108/2011 Rev. 1.
[7] ConsumerLab (2022). Pyrrolizidine alkaloids in supplements. View source
[8] American Herbal Products Association (2025). Code of Ethics, Business Conduct and Trade Requirements. View source
Reviewed by
Prof. Bingji MaScientific Advisor at PhytoVale. Professor and medicinal plant science expert. Research focus covers Chinese medicinal materials, phytochemistry, natural products and medicinal plant quality.
Written by
FloraContent Editor at PhytoVale. Product content and editorial specialist, focused on botanical ingredient specifications, sourcing information and B2B technical content.
This article is intended for ingredient sourcing, formulation and technical information only. It is not medical advice and makes no health claim for any PhytoVale product. The laboratory comparison and the upper-bound chart in section 03 use illustrative figures to explain how results are calculated and do not describe any real lot. Regulatory limits and positions summarised here change over time and should be verified against the legislation and guidance in force in the destination market at launch. Finished-product manufacturers remain responsible for regulatory status, permitted claims and labelling.
Need the Right Botanical Extract for Your Formula?
Tell us your application and quality requirements. We’ll recommend a suitable botanical extract grade and provide samples, COA, specifications and pricing.
Standardized botanical extracts with reliable quality documentation.
