Pyrrolizidine Alkaloids and Liver Safety

This is the article that decides whether the other three matter. Butterbur is not a herb with a vague theoretical concern attached; it is a herb whose raw material contains a class of compounds that destroy the small blood vessels of the liver, damage DNA, and cause cancer in animals. Those compounds are called pyrrolizidine alkaloids, and they are the reason a mainstream neurology guideline retired a Level A recommendation, the reason products were pulled from sale in the United Kingdom, and the reason no honest page about this plant can lead with its benefits.

Nothing here is intended to frighten you away from a decision you make with a doctor. It is intended to make sure the decision is informed, because the crucial fact about butterbur is one most marketing copy leaves out: the safety of a butterbur product is a property of the factory, not of the plant.

Table of Contents

1. What Pyrrolizidine Alkaloids Are
2. How They Injure the Liver
3. Veno-Occlusive Disease: What It Looks Like
4. DNA Damage and Cancer
5. How Much Is in Butterbur, and Where
6. What “PA-Free” Processing Does — and Cannot Guarantee
7. Liver Injury Reported With Standardised Products
8. What Regulators Actually Did
9. Pregnancy, Breastfeeding and Children
10. Other Sources That Add to the Same Total
11. A Practical Monitoring Plan
12. Who Must Never Take Butterbur
13. Key Research Papers
14. Connections

What Pyrrolizidine Alkaloids Are

Pyrrolizidine alkaloids are natural pesticides. Plants make them to deter insects and grazing animals, and more than six thousand plant species — perhaps three per cent of all flowering plants — produce them. They occur in ragwort, comfrey, coltsfoot, borage, heliotrope, some Senecio and Crotalaria species, and in Petasites, the genus butterbur belongs to.

Two structural facts do most of the work in understanding the risk:

• The dangerous ones are the 1,2-unsaturated pyrrolizidine alkaloids — those with a double bond in a particular position of the ring system. Saturated pyrrolizidine alkaloids are largely harmless. Unfortunately the ones found in the plants people actually consume are usually the unsaturated kind.
• Many occur as N-oxides, which are more water-soluble and were once assumed to be safer. They are not: gut bacteria and gut-wall enzymes convert N-oxides back into the parent alkaloid. This is why a water-based preparation such as a tea is not a way around the problem — it may in fact extract the N-oxides very efficiently.

The exposure that matters is cumulative. Pyrrolizidine alkaloid toxicity is not principally an allergic or idiosyncratic phenomenon where a rare unlucky person reacts to a single dose. It is a dose-and-time problem: small amounts, repeated over weeks or months, add up in the liver. That is exactly the pattern of a daily migraine preventive taken for four months, which is why the duration limits in these articles are not decoration.

How They Injure the Liver

Here is the mechanism, in plain terms, because understanding it explains every practical rule that follows.

Pyrrolizidine alkaloids as they come out of the plant are not themselves very toxic. They become toxic because of what your own liver does to them. Liver enzymes of the cytochrome P450 family — the same enzymes that metabolise most medicines — convert them into highly reactive intermediates known as dehydropyrrolizidine or pyrrolic esters. These are chemically aggressive electrophiles with a very short half-life, and they attack whatever is nearest.

Two kinds of damage follow:

1. Protein and cell damage. The reactive metabolites bind to proteins in the cells lining the smallest liver blood vessels — the sinusoidal endothelial cells. Those cells swell, lose their grip on the vessel wall, and detach. The debris blocks the sinusoids and the small hepatic venules downstream. Blood can no longer drain properly through the liver.
2. DNA damage. The same reactive species bind to DNA, forming adducts that can cause mutations if not repaired. This is the basis of the cancer concern, and it is why there is no comfortable “safe” threshold in the way there is for a drug that simply irritates the stomach.

Because the bioactivation happens in the liver itself, the liver takes the brunt. And because the process consumes protective glutathione, the injury is worse in a liver already under load — from alcohol, from other medicines, from existing disease, or from poor nutrition.

One consequence deserves emphasis. A conventional drug-induced liver injury usually announces itself: enzymes rise, you feel unwell, you stop the drug, you recover. Pyrrolizidine-alkaloid injury can be silent while it accumulates and present late, as an obstructive vascular problem rather than a simple hepatitis. Feeling fine is not evidence that nothing is happening.

Veno-Occlusive Disease: What It Looks Like

The characteristic pyrrolizidine-alkaloid injury has a name: hepatic veno-occlusive disease, now more often called sinusoidal obstruction syndrome. The small outflow vessels of the liver are blocked, blood backs up, and the liver becomes congested.

Typical features are a tender, enlarged liver; abdominal swelling from ascites (fluid in the abdomen); rapid weight gain from that fluid; jaundice; and nausea. Severe cases progress to liver failure and can be fatal. Milder cases may resolve, but repeated or continued exposure can lead to fibrosis and cirrhosis.

This is not a hypothetical mechanism. The medical literature contains documented cases of veno-occlusive disease caused by pyrrolizidine-alkaloid-containing plants, including:

• Poisoning from Senecio in the United States, reported in Gastroenterology in 1977.
• Herbal-tea-induced veno-occlusive disease in adults in Hong Kong, reported in The Lancet in 1983 and followed by a 1985 paper in Gut that actually quantified the alkaloid dose the patients had consumed.
• Reversible veno-occlusive disease in an infant after consumption of a pyrrolizidine-containing herbal tea, reported in the European Journal of Pediatrics in 1995.

An honest distinction: those cases came from other pyrrolizidine-alkaloid plants, not from butterbur. There is no series of butterbur-caused veno-occlusive disease. What the cases establish is the class hazard — that these compounds, taken repeatedly in herbal preparations, do this to human livers, including infant livers. Butterbur’s raw material contains compounds of that class. That is the chain of reasoning, and it is why the raw plant is off the table entirely rather than merely discouraged.

Beyond the liver, pyrrolizidine alkaloids have been implicated in pulmonary hypertension and in congenital abnormalities following exposure in pregnancy — which is the toxicological basis for the absolute contraindication in pregnancy set out below.

DNA Damage and Cancer

The genotoxicity is a separate problem from the liver injury, and in some ways a more troubling one, because it does not require you to feel unwell for harm to have occurred.

The reactive metabolites form DNA adducts. Adducts that escape repair can become permanent mutations. In animal studies, several pyrrolizidine alkaloids cause liver tumours, and the International Agency for Research on Cancer has classified one of them, riddelliine, as possibly carcinogenic to humans. Reviews of the field have argued that pyrrolizidine alkaloids may contribute to human liver cancer in populations with chronic dietary exposure.

The practical implication is uncomfortable but simple. For a compound that damages DNA, regulators generally assume there is no threshold below which exposure is entirely without risk — only levels at which the risk becomes negligibly small. That is why the limits set for pyrrolizidine alkaloids in medicines are measured in micrograms per day, and why “a small amount of butterbur can’t hurt” is the wrong frame. The right question is not “is this dose safe?” but “has this product been measured, and is the residue below the limit of detection?”

How Much Is in Butterbur, and Where

Petasites species produce a range of pyrrolizidine alkaloids — senecionine and senkirkine are among those identified in the genus — and content varies by plant part, growing site, season and genotype. Both the rhizome (the part used for the migraine extract) and the leaf (used for the allergy extract) carry them.

Analytical chemists have developed methods that measure the useful sesquiterpenes and the harmful alkaloids in the same run, in raw rhizome material and in marketed dietary supplements. That work is more important than it sounds. Before those methods existed, “PA-free” was an assertion. With them, it becomes a measurable property that a laboratory can confirm or refute — which means a manufacturer’s refusal to show you a test result is now a meaningful refusal.

The essential consequence for anyone at home: there is no preparation technique that removes pyrrolizidine alkaloids. Not boiling, not long steeping, not drying, not fermenting, not tincturing in alcohol, not adding lemon. Water-based preparations extract the N-oxides efficiently, and your gut converts those back to the parent alkaloids. Removing pyrrolizidine alkaloids requires selective industrial extraction followed by analytical verification. A butterbur tea is not a weaker version of the medicine; it is the dangerous part of the plant with the useful part largely left behind.

What “PA-Free” Processing Does — and Cannot Guarantee

The standardised extracts exploit a difference in chemistry. The petasins are relatively non-polar, oily molecules; the pyrrolizidine alkaloids are more polar. A supercritical carbon-dioxide extraction — carbon dioxide held under pressure until it behaves like a solvent — preferentially dissolves the non-polar petasins and leaves most of the alkaloids behind. Further purification steps follow, and reputable manufacturers then specify the residual alkaloid content as being below the analytical limit of detection, batch by batch.

That is a genuine piece of engineering and it works. Three limits on how much comfort it should give you:

1. “Below the limit of detection” is not zero. It is a statement about the instrument. A more sensitive method finds more. For a genotoxic compound taken daily for months, the difference is not purely academic.
2. It is a batch property, not a plant property. Every batch has to be extracted correctly and tested. A process that works is not the same as a process that worked on the bottle in your hand.
3. Nobody checks before sale in the United States. Butterbur is sold there as a dietary supplement, which means the manufacturer is responsible for its own claims and no agency verifies alkaloid content before the product reaches a shelf. The claim on the label and the contents of the capsule are connected only by the manufacturer’s own diligence.

This is why Choosing a PA-Free Product exists as a separate article. With most herbs, product quality affects whether you get an effect. With butterbur, product quality is the entire safety margin.

Liver Injury Reported With Standardised Products

If the processing removes the alkaloids, why is anyone still worried? Because liver injuries have been reported in people taking commercial, supposedly PA-free butterbur.

Reports have described raised liver enzymes and, in some cases, hepatitis in users of butterbur supplements, including the standardised migraine extract. Anderson and Borlak examined these hepatobiliary events in detail in the Journal of Clinical Medicine in 2019, working through the published and reported cases. Their analysis concluded that the case material was largely too poorly documented — confounded by other medications, other illnesses, missing timelines — to establish that the extract itself caused the injuries. A 2022 narrative review co-authored by the same investigator, with prominent headache researchers, argued the same position more broadly.

Read that alongside three other facts, and form your own view:

• Regulators looking at the same signal acted anyway, restricting or removing products in more than one country.
• The authors arguing the signal is weak have published repeatedly in defence of this extract. That does not make them wrong; it is context you are entitled to.
• “Causality not established” is not the same as “causality excluded.” Spontaneous adverse-event reports are almost always poorly documented; that is a known weakness of the reporting system, not evidence of innocence.

The honest state of the field is a genuine, unresolved disagreement between people who think the case reports reflect residual alkaloids or an idiosyncratic reaction to some other constituent, and people who think they reflect confounding and coincidence. When experts disagree about whether a liver risk is real, a patient with a good alternative should generally take the alternative.

What Regulators Actually Did

Regulatory history is the clearest evidence of how seriously this was taken by people with access to the underlying safety data.

• Germany. German medicines regulators have applied strict ceilings on pyrrolizidine-alkaloid intake from internally used medicinal products since the early 1990s — limits in the micrograms per day range, with a substantially lower ceiling for use beyond a few weeks a year. The marketing authorisation of the leading standardised butterbur extract was withdrawn in Germany over hepatic safety concerns.
• United Kingdom. In 2012 the medicines regulator warned the public about unlicensed butterbur products following reports of liver toxicity, and such products were withdrawn from sale.
• European Union. Maximum levels for pyrrolizidine alkaloids have since been set for herbal infusions and for food supplements containing botanical ingredients — an acknowledgement that this is a food-chain problem, not only a herbal-medicine one.
• United States. Butterbur remains available as a dietary supplement. There is no pre-market verification of alkaloid content. The National Center for Complementary and Integrative Health advises that only products certified and labelled as PA-free should be considered.
• American Academy of Neurology. The 2012 guideline that rated Petasites Level A for migraine prevention now carries a [RETIRED] designation, and in 2015 the Academy acted to make clinicians and patients aware of the hepatotoxicity concern. The efficacy data did not change; the safety judgement did.

Notice the shape of that record. This is not a case of an effective natural remedy being suppressed. Regulators had a product that worked and they still restricted it, because they could not be confident about what was in each bottle.

Pregnancy, Breastfeeding and Children

These are absolute contraindications, and it is worth saying why rather than simply asserting it.

Pregnancy. Pyrrolizidine alkaloids cross the placenta. The foetal liver is developing, the exposure is to a genotoxic compound, and pyrrolizidine alkaloids have been implicated in congenital abnormalities. There is no version of this risk-benefit calculation that favours a migraine or hay-fever supplement. Migraine frequently improves during pregnancy in any case, and pregnancy-safe options for both conditions exist and are well characterised — ask your obstetric team.

Breastfeeding. Pyrrolizidine alkaloids pass into breast milk. An infant’s liver has limited capacity and the documented infant case of herbal-tea veno-occlusive disease shows what the consequences can look like. Do not take butterbur while breastfeeding.

Children. Three arguments converge. The paediatric efficacy evidence is an uncontrolled open study, one small explorative trial and a retrospective chart review — not an evidence base. A genotoxic exposure matters more when there are seventy years of life left for a mutation to express itself. And a child cannot weigh the trade-off or report early symptoms reliably. Childhood migraine and childhood hay fever both have better-supported, safer treatments.

Other Sources That Add to the Same Total

Because the risk is cumulative, what matters is your total pyrrolizidine-alkaloid intake, not the contribution of any one product. People often carry several sources without realising it.

• Other PA-containing herbs: comfrey, coltsfoot, borage, ragwort, heliotrope, and various traditional Senecio and Crotalaria preparations. Comfrey in particular is still sold and still recommended for internal use in some circles; it should not be taken internally at all.
• Contaminated herbal teas and honey. Pyrrolizidine alkaloids reach the food chain when PA-producing weeds are harvested alongside crops such as chamomile, rooibos or peppermint, or when bees forage on ragwort. This is precisely why the European Union set maximum levels.
• Anything that loads the liver in parallel: alcohol, methotrexate, isoniazid, amiodarone, ketoconazole, frequent or high-dose paracetamol/acetaminophen, and high-dose niacin. None of these interacts with butterbur in a pharmacokinetic sense; they simply add injury to the same organ.
• Existing liver disease — fatty liver disease, hepatitis B or C, prior drug-induced injury, cirrhosis — reduces the reserve you have to absorb any additional insult.

A Practical Monitoring Plan

If you and a clinician decide to proceed with a certified PA-free standardised product, this is the minimum sensible structure. It is not a substitute for that clinician’s advice.

Stop immediately and seek medical care for

• Yellowing of the skin or the whites of the eyes
• Dark, tea-coloured urine, or pale, clay-coloured stools
• Persistent nausea, vomiting or loss of appetite
• Pain or tenderness under the right ribs
• Abdominal swelling or unexplained rapid weight gain
• Unexplained itching
• Unusual, profound fatigue

Tell whoever sees you that you have been taking butterbur, and take the bottle with you. Herbal supplements are a well-recognised cause of liver injury and are frequently missed because patients do not think of them as medicines.

Who Must Never Take Butterbur

• Anyone who is pregnant, trying to conceive, or breastfeeding
• Children and adolescents
• Anyone with liver disease of any kind, or a history of drug-induced liver injury
• Anyone drinking alcohol heavily or regularly
• Anyone taking other hepatotoxic medicines without a clinician actively supervising the combination
• Anyone with a known allergy to Asteraceae plants — ragweed, chrysanthemum, marigold, daisy, chamomile, echinacea, feverfew
• Anyone using the raw herb, a wild-harvested plant, a home-made tea or tincture, or bulk powder of unknown origin — without exception
• Anyone who cannot verify that a specific product has been tested and certified free of pyrrolizidine alkaloids

And a closing thought that belongs on this page more than any other. “Natural” describes where a molecule came from, not what it does. Pyrrolizidine alkaloids are entirely natural — they are a plant’s chemical defence system, evolved specifically to poison animals that eat it. Working exactly as intended is the problem, not a malfunction.

Back to Table of Contents

Key Research Papers

Every identifier below was checked live against the NCBI PubMed database — title, first author, journal and year all had to match. Note carefully which papers concern butterbur specifically and which concern the alkaloid class as a whole; that distinction is made explicit in each entry.

Butterbur-specific safety

1. Anderson N, Borlak J. Hepatobiliary events in migraine therapy with herbs — the case of Petadolex, a Petasites hybridus extract. Journal of Clinical Medicine. 2019;8(5). Detailed analysis of reported liver events with the standardised migraine extract; concluded the case material was largely too poorly documented to establish causality.
2. Borlak J, Diener HC, Kleeberg-Hartmann J, Messlinger K, Silberstein S. Petasites for migraine prevention: new data on mode of action, pharmacology and safety. A narrative review. Frontiers in Neurology. 2022;13:864689. Narrative review arguing the hepatic signal is weaker than assumed — a position to weigh against the regulatory record, not a substitute for it.
3. Sutherland A, Sweet BV. Butterbur: an alternative therapy for migraine prevention. American Journal of Health-System Pharmacy. 2010;67(9):705–711. Pharmacy review setting the efficacy data against the alkaloid problem.
4. Aydın AA, Zerbes V, Parlar H, Letzel T. The medical plant butterbur (Petasites): analytical and physiological (re)view. Journal of Pharmaceutical and Biomedical Analysis. 2013;75:220–229. Review of the plant’s chemistry with the analytical problem at its centre.

Measuring the alkaloids in butterbur and in products

1. Avula B, Wang YH, Wang M, Smillie TJ, Khan IA. Simultaneous determination of sesquiterpenes and pyrrolizidine alkaloids from the rhizomes of Petasites hybridus (L.) G.M. et Sch. and dietary supplements using UPLC-UV and HPLC-TOF-MS methods. Journal of Pharmaceutical and Biomedical Analysis. 2012;70:53–63. Analytical chemistry applied to raw rhizome material and to marketed supplements — the work that makes an independent “PA-free” check possible.
2. Aydın AA, Letzel T. Simultaneous investigation of sesquiterpenes, pyrrolizidine alkaloids and N-oxides in butterbur (Petasites hybridus) with an offline 2D-combination of HPLC-UV and LC-MMI-ToF-MS. Journal of Pharmaceutical and Biomedical Analysis. 2013;85:74–82. Includes the N-oxides, which the gut converts back to the parent alkaloids.

The alkaloid class: toxicology and carcinogenicity

1. Edgar JA, Molyneux RJ, Colegate SM. Pyrrolizidine alkaloids: potential role in the etiology of cancers, pulmonary hypertension, congenital anomalies, and liver disease. Chemical Research in Toxicology. 2015;28(1):4–20. The case that chronic low-level exposure matters, not only acute poisoning.
2. Neuman MG, Cohen L, Opris M, Nanau RM, Hyunjin J. Hepatotoxicity of pyrrolizidine alkaloids. Journal of Pharmacy & Pharmaceutical Sciences. 2015;18(4):825–843. Review of the bioactivation and sinusoidal-obstruction mechanism.
3. Moreira R, Pereira DM, Valentão P, Andrade PB. Pyrrolizidine alkaloids: chemistry, pharmacology, toxicology and food safety. International Journal of Molecular Sciences. 2018;19(6). Covers the food-chain contamination route as well as medicinal exposure.

Human veno-occlusive disease from PA-containing plants (not butterbur)

1. Stillman AS, Huxtable R, Consroe P, Kohnen P, Smith S. Hepatic veno-occlusive disease due to pyrrolizidine (Senecio) poisoning in Arizona. Gastroenterology. 1977;73(2):349–352. Human case report.
2. Kumana CR, Ng M, Lin HJ, Ko W, Wu PC, Todd D. Hepatic veno-occlusive disease due to toxic alkaloid herbal tea. The Lancet. 1983;2(8363):1360–1361. Human cases from a herbal tea.
3. Kumana CR, Ng M, Lin HJ, Ko W, Wu PC, Todd D. Herbal tea induced hepatic veno-occlusive disease: quantification of toxic alkaloid exposure in adults. Gut. 1985;26(1):101–104. The follow-up that measured how much alkaloid the patients had actually consumed.
4. Sperl W, Stuppner H, Gassner I, Judmaier W, Dietze O, Vogel W. Reversible hepatic veno-occlusive disease in an infant after consumption of pyrrolizidine-containing herbal tea. European Journal of Pediatrics. 1995;154(2):112–116. Human infant case — the clearest reason these plants are contraindicated in children and in breastfeeding.

Live PubMed Searches

1. Pyrrolizidine alkaloid hepatotoxicity
2. Sinusoidal obstruction syndrome
3. PA DNA adducts and genotoxicity
4. Petasites and pyrrolizidine alkaloids
5. Petadolex and hepatotoxicity
6. Herbal supplement induced liver injury
7. PA contamination of teas and honey
8. Comfrey, pyrrolizidine alkaloids and the liver

Connections

• Butterbur Benefits — hub
• Choosing a PA-Free Product
• Migraine Prevention
• Allergic Rhinitis and Hay Fever
• Butterbur
• Comfrey — the other pyrrolizidine-alkaloid herb
• Liver Function Tests
• GGT
• Milk Thistle
• Toxins
• Gastroenterology
• Neurology
• Feverfew
• All Herbs

Back to Table of Contents