Comfrey

What is Comfrey?

Comfrey is Symphytum officinale L., and the FDA's action against oral comfrey is a letter dated 6 July 2001, not a regulation. That distinction runs through this issue. The letter, signed by Christine J. Lewis of the Office of Nutritional Products, Labeling, and Dietary Supplements, went to eight trade associations and said the agency "strongly recommends that firms marketing a product containing comfrey or another source of pyrrolizidine alkaloids remove the product from the market and alert its customers to immediately stop using the product." It stated FDA's belief that such products are adulterated. It created no rule. Twenty-five years later the NIH label database lists comfrey leaf capsules at 450 mg, bulk comfrey leaf powder and both leaf and root tinctures as still on sale to US buyers.

The plant is a Boraginaceae perennial with a thick mucilaginous taproot, naturalized across North America. Its reputation rests on allantoin, 0.6 to 4.7 percent of the root, and on pyrrolizidine alkaloids, up to 8,320 micrograms per gram of root. The topical story that follows is genuinely positive and genuinely narrow.

Common Names

• Comfrey, common comfrey, knitbone, boneset, bruisewort, slippery root, Beinwell in German, consoude in French

• Symphytum officinale L. is the true species. Symphytum asperum Lepech. is prickly comfrey, and Symphytum x uplandicum Nyman, the hybrid of the two, is Russian comfrey, also sold as Symphytum peregrinum

• The hybrid and S. asperum carry echimidine; S. officinale does not, which makes echimidine and symlandine the standard markers for substitution of the hybrid into material labeled S. officinale

• Hirono's 1978 carcinogenicity paper is titled "Carcinogenic activity of Symphytum officinale" and calls the material "Russian comfrey" in its own abstract, so the species behind the most-cited animal study is ambiguous

• Not Digitalis purpurea. Foxglove leaves resemble comfrey leaves out of bloom, and at least four published poisoning incidents followed that mistake, one of them fatal

Primary Active Compounds

• Allantoin, 0.6 to 4.7 percent of the root on Staiger's compilation, the constituent behind the tissue-repair claim

• Mucilage polysaccharides, about 29 percent of the root, built from fructose and glucose units

• Rosmarinic acid up to 0.2 percent, chlorogenic acid 0.012 percent, caffeic acid 0.004 percent, all three from Staiger where the EMA assessment report gives higher acid figures, plus globoidnan A, globoidnan B and rabdosiin, proposed in 2020 as quality markers

• Triterpene saponins on hederagenin and oleanolic acid aglycones, including symphytoxide A, plus a glycopeptide fraction

• Pyrrolizidine alkaloids, almost entirely as N-oxides: 7-acetylintermedine and 7-acetyllycopsamine dominate, with smaller intermedine, lycopsamine and symphytine

Key Note

The alkaloids are not a contaminant. They are a constitutive product of the plant's own secondary metabolism, concentrated in the root, and no amount of clean sourcing removes them. The only comfrey material that reliably lacks them is a clonally propagated cultivar bred for their absence.

Comfrey holds a real topical effect and a documented oral hepatotoxin in the same tissue, and most products blur which one you are buying.

Primary Functions & Benefits

Topical anti-inflammatory and analgesic effect, the one function with controlled human data

Four controlled trials of the 35 percent root ointment and three of the 10 percent herb cream report significant pain or wound-area improvement; effect sizes and denominators are in Benefits of Taking Comfrey. Every one was cutaneous. Staiger's 2012 review in Phytotherapy Research adds that comfrey "stimulates granulation and tissue regeneration, and supports callus formation," while conceding in the same paragraph that the activity-determining constituents and the molecular mechanism "have not been completely elucidated." No trial has measured callus formation or fracture healing in people, which is what "knitbone" implies.

Hepatotoxicity and hepatocarcinogenicity, a function of the plant and not a benefit

Hirono, Mori and Haga fed inbred ACI rats comfrey leaves for 480 to 600 days and comfrey roots for 245 or 280 days or until death (Journal of the National Cancer Institute 1978, volume 61, pages 865 to 869). Hepatocellular adenomas appeared in all nine comfrey groups: across four leaf groups, 5 of 19 at 33 percent of the diet over 480 days and 11 of 20 at the same 33 percent over 600, then 7 of 21 at 16 percent and 1 of 28 at 8 percent; across five root groups, 19 of 24 at 8 percent down to 7 of 24 on a tapering schedule. Controls were 0 of 129, and the paper gave no significance tests. Hemangioendothelial sarcoma of the liver appeared infrequently. Separately, Hirono and colleagues in 1979 gave symphytine intraperitoneally over 56 weeks and found liver tumors in 4 of 20 rats against 0 of 20, which the NTP review calls suggestive but not statistically significant. No human pharmacokinetic study of comfrey extract constituents has been published, by Staiger's own account in a review she wrote as a Merck Selbstmedikation employee.

Forms & Standardization

The forms differ in alkaloid load by five orders of magnitude, the comparison that decides whether a comfrey product is a medicine or a liver toxin.

Kytta-Salbe f, the 35 percent root ointment, the most-tested preparation

35 g per 100 g of a liquid extract of fresh S. officinale root, drug-to-extract ratio 1:2, extraction solvent ethanol 60 percent by volume. Made by Merck Selbstmedikation. This is the material in the four trials named above, plus a 306-child observational series in ages 3 to 12.

Its alkaloid ceiling appears at "less than 0.35 ppm" in Staiger 2012 and in the EMA draft assessment report's description of the trial ointment, and at "less than 35 ppm" in the final assessment report as the end-product specification, a hundredfold apart. The draft report also gives "less than 1 ppm" for extract batches used in the non-clinical work, which is a different thing again. All are maxima, not measurements, and no source states which alkaloids the figure sums.

Traumaplant, the 10 percent herb cream built on a low-alkaloid clone

1 g of preparation per 10 g of cream, made of 0.4 g pressed plant juice and 0.6 g 30 percent ethanol extract from the aerial parts of Symphytum x uplandicum Nyman cultivar 'Harras', equivalent to about 25 g of fresh herb per 100 g of cream. Harras Pharma Curarina, Munich. Kucera and colleagues reported in 2018 that "potentially toxic pyrrolizidine alkaloids were not present in either the active substance concentrate or the cream preparation (detection limit less than 0.01 micrograms per gram)." That is a limit of detection, not a measured zero, and the plant is the hybrid, which carries echimidine in wild material.

The EU monograph preparation, a third thing again

European Union herbal monograph EMA/HMPC/572846/2009 Corr1, adopted 5 May 2015, covers a liquid extract of S. officinale root in ethanol 65 percent by volume, partially evaporated and adjusted to drug-to-extract ratio 2:1, formulated at 10 percent in a semi-solid. That is neither Kytta-Salbe f nor Traumaplant: Europe's regulatory benchmark describes a preparation no major trial used.

Dried root and leaf, cut or powdered, the highest-alkaloid forms sold

Couet and colleagues measured 1,380 to 8,320 micrograms of total pyrrolizidine alkaloid per gram of root in commercial S. officinale samples, against 15 to 55 micrograms per gram of leaf. The NTP background document puts dry root at 0.2 to 0.4 percent, 2,000 to 4,000 micrograms per gram, and dry leaf at 0.003 to 0.2 percent, 30 to 2,000. On the one dataset that measured both organs in the same commercial samples, root runs 25-fold to 555-fold above leaf, about 140-fold at the midpoints.

Capsules, teas, and "comfrey extract" at unstated strength

In the NIH Dietary Supplement Label Database the only single-herb comfrey capsule on the US market is a 450 mg leaf capsule, the only single-herb powder is leaf, and the root products are tinctures, one declared as dried Symphytum officinale root liquid extract. Bulk root sold as a craft or garden herb rather than a labeled supplement sits outside that database, so absence there is not absence from the market. The UK Food Standards Agency measured one retail comfrey infusion at 52,508 micrograms of total alkaloid per kilogram, 52.5 micrograms per gram, dominated by intermedine and lycopsamine. None declares a species beyond "comfrey" or a cultivar. Their arithmetic is in Toxicity Symptoms, as a hazard calculation and not a dose.

Food Sources

Comfrey is not a food, so this is routes of exposure, and the dominant one is ingestion.

Deliberate oral intake, the route that produced every human case

Leaf tea, root tea, root powder, capsules and tincture. Every case in Toxicity Symptoms traces to this route. The leaves were historically eaten as a cooked green, which is how Hirono's rats got theirs.

Topical application, the route with a measured absorbed fraction

Through intact human skin the absorbed fraction is 0.04 to 4.9 percent of what is applied, worked out in Recommended Dosages.

Misidentification, which has produced more acute harm than comfrey itself

Lin and colleagues reported an outbreak of nine patients poisoned by an alleged comfrey herbal tea that was Digitalis purpurea (Journal of the Chinese Medical Association 2010, volume 73, pages 97 to 100). Peak serum digoxin ran 4.4 to 139.5 ng/mL, three patients developed significant cardiotoxicity with mild hyperkalemia and needed temporary cardiac pacing, and 40 to 80 mg of digoxin-specific antibody had no effect. All nine recovered. Vithayathil and Edwards described a 63-year-old woman who boiled "comfrey" leaves for insomnia and reached a London emergency department 18 hours later with second-degree atrioventricular block, Mobitz Type 2, and a raised serum digoxin (BMJ Case Reports 2016); she survived. The mistake has also killed. Clinical Toxicology published a fatal cardiac glycoside poisoning from foxglove taken for comfrey in 2017 (volume 55, issue 7), and a further Digitalis purpurea tea poisoning was reported in 2021.

Dietary background from other plants

Pyrrolizidine alkaloids reach food from unrelated weeds, which is why the European Union set maximum levels by Regulation (EU) 2020/2040, applying from 1 July 2022, on the sum of 35 alkaloids in teas, infusions, herbal supplements, pollen products and some spices. The UK FSA survey that measured the 52,508 micrograms per kilogram comfrey infusion found its highest other plant supplement at 344 micrograms per kilogram, a marshmallow capsule product. Comfrey is the outlier by a factor of 150.

Is dietary intake realistically sufficient?

There is no comfrey requirement and nothing for food to supply. Eat the food and skip the pill does not apply either, because the food was the hazard: Hirono's adenomas came from comfrey leaves in the diet.

Who Should Take Comfrey

Nobody should take comfrey by mouth, in any form, at any dose. The FDA's 2001 request, AHPA's trade requirement and the European monographs all point the same way. What follows is who has a topical indication and who should think about exposure.

Who has a real topical indication

• Acute ankle sprain within hours of injury, with either preparation

• Acute upper or lower back pain, over 5 days with the root ointment and 8 to 10 with the herb cream

• Painful knee osteoarthritis, over 21 days with the root ointment

• Fresh abrasions on otherwise intact surrounding skin, with the 10 percent herb cream

• Children aged 4 and over, for the herb cream, which is the floor its pediatric studies set. German national labels for the 35 percent root ointment cover age 3 and over, while the EU monograph for root preparations does not recommend use under 18, a more restrictive position on the same plant part

Who is unavoidably exposed

Gardeners and smallholders who grow comfrey as a compost activator or fodder and handle it bare-handed: raw plant material carries an alkaloid load two to four orders of magnitude above a labeled cream, depending on plant part.

Who should consider testing

There is no routine clinical test for comfrey exposure. A reader with months of oral comfrey behind them and new abdominal distension or right upper quadrant pain needs liver function tests and imaging for ascites. Pyrrolizidine protein adduct assays are research-only.

Who Should AVOID or Use Caution

Contraindications

• Oral use by anyone. This is absolute

• Pregnancy and breastfeeding. No topical comfrey product carries a pregnancy indication, and AHPA's required label statement says "Do not use when nursing" in plain words

• Broken, abraded or irritated skin, for S. officinale root preparations. The EU monograph forbids it, and every dermal absorption figure came from intact skin

• Existing liver disease of any cause, including hepatitis B or C, alcohol-related liver disease and fatty liver. The target cell in pyrrolizidine injury is the hepatic sinusoidal endothelial cell, and a liver already carrying sinusoidal pathology has less margin

• Eyes and mucous membranes

Use Caution

• Children under 4 for the herb cream and under 3 for a root product, the floors their own studies and labels set, and under 18 for root preparations where the EU monograph draws its line

• Any use beyond 10 consecutive days with a root preparation, or over large surface areas. The absorption arithmetic assumes 6 to 12 g a day on one joint, not a limb

• Products labeled only "comfrey," with no species, plant part, extract ratio or alkaloid figure, which is most US retail product, and homemade poultices, oils and salves from garden plants, where the load is unmeasured and the species is usually the hybrid

Critical Safety Point

The injury from oral comfrey is hepatic sinusoidal obstruction syndrome, and jaundice follows rather than opens it. LiverTox records that it emerges within 1 to 2 months of starting the product and presents with right upper quadrant pain, nausea and weight gain from fluid retention, followed by jaundice, with aminotransferases "usually only mildly elevated with a hepatocellular pattern of injury, although they may be markedly increased if tested during the early phases." A normal liver panel does not rule it out, and neither does a markedly abnormal one. The chronic form, which is what the long-term tea cases were, develops insidiously with ascites and weakness. LiverTox assigns comfrey a likelihood score of C, "a probable cause of clinically apparent liver injury due to sinusoidal obstruction syndrome," when taken orally.

Recommended Dosages

There is no recommended oral dose of comfrey: oral use is withdrawn and no number of milligrams is safe to swallow. Everything below is cutaneous, and specific on purpose.

35 percent root ointment (Kytta-Salbe f type), by indication, as the trials ran it

• Ankle sprain: a strand of ointment about 6 cm long, roughly 2 g, four times a day for 7 to 8 days, about 8 g a day. Koll 2004 and Predel 2005

• Acute back pain: 4 g three times a day for 5 days, about 12 g a day. Giannetti 2010

• Knee osteoarthritis: 2 g three times a day, 6 g a day, for 21 days. Grube 2007. That duration is 2.1 times the EU monograph's maximum and is a trial protocol, not a labeled regimen

EU monograph preparation, 10 percent of a 2:1 root extract in a semi-solid

A thin layer twice daily to intact skin, not more than 10 days, not recommended under 18 years. EMA/HMPC/572846/2009 Corr1, adopted 5 May 2015, as a traditional-use registration resting on long-standing use rather than on trials. German national labels for the root ointment are the other non-protocol dosing, at 2 to 4 times daily and 1.2 to 6 g.

10 percent herb cream on the low-alkaloid cultivar (Traumaplant type)

Two to four applications a day, the frequency the product's own guidance calls typical, for up to 14 days. The 2018 series in 712 children recorded one to five a day, which is what patients did and not what the label asks. The abrasion trials used it on fresh wounds. It is the only comfrey preparation with published broken-skin data, and the lowest measured alkaloid load.

Duration

Ten consecutive days per course for root preparations, the EU monograph ceiling. Germany's 1992 Bundesgesundheitsamt Stufenplan decision allows 100 micrograms of 1,2-unsaturated alkaloid per day externally for at most six weeks a year, and no duration limit at or below 10 micrograms a day. A labeled ointment falls under the lower figure, so the binding constraint on it is the monograph's 10 days.

The applied dose, step by step

Start from the tightest published specification for the 35 percent root ointment, under 0.35 micrograms per gram of finished ointment. Giannetti's 12 g a day gives 12 x 0.35 = 4.2 micrograms applied per day; Predel and Koll's roughly 8 g a day gives 2.8; Grube's 6 g a day gives 2.1. Giannetti's regimen is 4.2 percent of Germany's 100 microgram external ceiling and 42 percent of the 10 microgram no-duration-limit threshold.

At the loosest published specification for the same extract, under 35 micrograms per gram, Giannetti's 12 g a day becomes 420 micrograms applied, 4.2 times Germany's external ceiling. Which published number is right decides whether the most-cited comfrey trial complied with the governing limit, and the record does not resolve it. For the Traumaplant-type cream, 10 g a day is under 0.1 micrograms applied, on a detection limit rather than a measured content.

The absorbed dose, step by step

Kuchta and Schmidt applied spiked comfrey cream to human abdominal skin in Franz diffusion cells (Regulatory Toxicology and Pharmacology 2020, volume 118). In five of six cells no lycopsamine was detectable in the skin, with 0.6 plus or minus 0.4 percent of the applied dose in the receptor fluid, and their worst-case penetration estimate was 4.9 percent. On human epidermis, a different model, Jedlinszki and colleagues measured 0.04 to 0.22 percent over 24 hours, the figure the EMA addendum carries. Both authors have herbal-industry ties and their paper argues the limits overstate topical risk, so read 4.9 percent as the figure they conceded. Applied to Giannetti's 4.2 micrograms a day, that gives 0.0017 micrograms absorbed at 0.04 percent, 0.025 at 0.6 percent, and 0.21 at 4.9 percent.

The reference point is 1.0 micrograms a day of 1,2-unsaturated alkaloid for an adult, from EMA/HMPC/893108/2011 Rev. 1, adopted 7 July 2021 and applying to cutaneous use as well as oral. It derives from riddelliine, not from comfrey's own alkaloids: EFSA's 2017 benchmark dose lower confidence limit of 237 micrograms per kilogram body weight per day for liver hemangiosarcoma in female rats, divided by 10,000 under ICH M7, giving an acceptable intake of 0.0237 micrograms per kilogram body weight. At the HMPC's assumed 50 kg adult, not 70, that is 1.185, rounded to 1.0. The superseded 2014 figure of 0.35 came another way, a TD50 calculation on lasiocarpine with a factor of 50,000. The history runs opposite to the obvious reading: in 2016 the EMA made 1.0 a three-year transitional ceiling and 0.35 the target producers were to work down to, extended that in 2019, and only in Rev. 1 did 1.0 become the standing limit. Addendum EMA/HMPC/313662/2023, adopted 31 January 2024, corrected the comfrey monograph to 1.0 and otherwise concluded no revision was needed. So the margin for a labeled topical course is 4.8-fold at the worst-case absorption estimate, 40-fold at the central one, 590-fold at the lowest.

Where the topical arithmetic turns against you

A poultice of 50 g of fresh comfrey leaf is roughly 7 g dry, at the 13.4 percent dry matter Feedipedia reports for fresh comfrey aerial parts. At Couet's leaf range of 15 to 55 micrograms per gram, that poultice puts 105 to 385 micrograms against skin, 25 to 92 times Giannetti's 4.2 micrograms. At 0.6 percent absorption that is 0.6 to 2.3 micrograms a day; at 4.9 percent, 5.1 to 18.9. The poultice crosses the 1.0 microgram reference at the central estimate and clears it severalfold at the upper one, where no labeled ointment does at any estimate. At the NTP document's upper dry-leaf figure of 2,000 micrograms per gram the same poultice carries 14,000. And a poultice usually goes on broken skin, where the absorption measurements do not apply at all.

Timing & Administration

Start as early after injury as you can, and apply often

Koll 2004 enrolled only patients whose ankle sprain had happened within the previous 6 hours, and that inclusion criterion is part of the result. The large sprain effects came from treating within hours, not from starting on day three. Frequency was three to four applications a day in every positive trial, 6 to 12 g of ointment a day; once-daily use was never tested.

Apply to intact skin for root preparations, and do not occlude

The EU monograph specifies a thin layer, less than the trials used. Keep a bandage off a root ointment: occlusion increases percutaneous absorption and the published dermal fractions were measured without it. Wash your hands afterward, because the practical oral exposure risk from a cream is hand-to-mouth transfer.

Timing relative to food and time of day, and use with an analgesic

Food and time of day are irrelevant for a cutaneous product with no measured human pharmacokinetics. Predel 2005 and Koll 2004 allowed paracetamol as rescue medication and counted its use as a secondary endpoint, so the ointment was tested alongside an oral analgesic, not instead of one.

Timeline of Effects

1 hour, back pain

Giannetti 2010 reported a measurable effect on pain during active standardized movement at 1 hour, which the paper calls the first demonstration of a fast-acting effect for the ointment. One trial, one endpoint, 120 patients.

Day 3 to day 4, sprain and abrasion

Barna 2007 found the faster wound-size reduction emerged after 2 to 3 days. Kucera 2004 found significant separation on days 3, 4 and 7 in ankle distortion, swelling on days 3 and 4.

Day 5 to day 8, the full sprain and back pain course

Giannetti ran 5 days, Koll 8, Predel 7 plus or minus 1. If a sprain or an acute back episode has not responded inside this window, the trials give you no reason to continue.

Day 21, knee osteoarthritis

Grube 2007 ran 21 days and reported that the gap between groups widened systematically with duration, which argues against stopping early.

When to conclude it is not working

Acute sprain or back pain with no change by day 5. Knee osteoarthritis with no change by day 21. Any skin reaction at the site. A fair trial means three applications a day minimum.

Benefits of Taking Comfrey

There is no benefit from swallowing comfrey; the oral history is handled as history below. The topical benefits are real.

Ankle sprain, against placebo and against diclofenac gel

Koll and colleagues randomized 142 patients with unilateral ankle sprain of under 6 hours' duration, mean age 31.8 years, 78.9 percent male, to the 35 percent root ointment or placebo four times daily for 8 days (Phytomedicine 2004, volume 11, pages 470 to 477). Pain on tonometry favored comfrey at p less than 0.0001 and edema by the figure-of-eight method at p = 0.0001, with no adverse drug reactions. The patient-reported visual analog scale scores for pain at rest and on movement showed no significant group difference.

Predel and colleagues then randomized 164 patients, 82 per arm, to the same ointment or a diclofenac gel containing 1.16 g of diclofenac diethylamine salt, four times daily for 7 plus or minus 1 days (Phytomedicine 2005, volume 12, pages 707 to 714). The design was non-inferiority. The 95 percent confidence interval for the difference in pressure-pain area under the curve was 19.08 to 103.09 h x N/cm2, entirely above the non-inferiority margin, so non-inferiority was confirmed and the authors noted the result "may be" superiority. D'Anchise, Bulitta and Giannetti reanalyzed the same dataset for superiority in 2007 and reported a mean difference of 61.1 h x N/cm2. That is one trial reported twice, not two trials.

Acute back pain

Giannetti and colleagues randomized 120 patients, mean age 36.9 years, to the 35 percent root ointment or placebo, 4 g three times daily for 5 plus or minus 1 days, at the German Sport University in Cologne and three orthopedic centers (British Journal of Sports Medicine 2010, volume 44, pages 637 to 641). Median reduction in pain on active standardized movement was 95.2 percent against 37.8. The 2014 Cochrane review of herbal medicine for low back pain graded this trial's evidence low quality.

Knee osteoarthritis

Grube and colleagues randomized 220 patients, 153 women and 67 men, mean age 57.9 years, with knee complaints of 6.5 years' standing, to 6 g a day of the 35 percent root ointment or placebo for 21 days (Phytomedicine 2007, volume 14, pages 2 to 10). All 220 were in the full analysis set; 186, or 84.5 percent, qualified for the valid case analysis. The visual analog scale total score fell 51.6 mm, or 54.7 percent, on comfrey against 10.1 mm, or 10.7 percent, on placebo, a difference of 41.5 mm, 95 percent confidence interval 34.8 to 48.2 mm, p less than 0.001. WOMAC fell 60.4 mm against 14.7. Those figures belong to the 220, the set the paper reports them in.

Fresh abrasions, including on children

Barna and colleagues randomized 278 patients with fresh abrasions, 137 to the 10 percent herb cream and 141 to an otherwise identical 1 percent cream. Initial wound-size reduction was 49 plus or minus 19 percent per day against 29 plus or minus 13, and linear regression put complete healing 2.97 days earlier, 4.08 against 7.05 days. Kucera's 215-patient back pain trial with the same cream, 104 against 111 on the 1 percent reference, gave a number needed to treat of 3.2.

The design caveat on every herb-cream trial

Kucera's and Barna's comparator was not a placebo. It was the same cream at 1 percent, 2.5 g of fresh herb per 100 g against 25 g. A published comment defended that as reasonable when blinding a herbal is difficult. It is defensible and it is also a dose comparison: those trials establish that more comfrey beats less, not that comfrey beats nothing.

The oral history, as history

"Knitbone" and "boneset" record a European practice of splinting fractures with comfrey root paste, which hardens as the mucilage dries: mechanical, not pharmacological. Comfrey tea and root tablets were sold for gastric and duodenal ulcers, colitis and arthritis into the 1980s. Commission E granted an external indication only.

Potential Negatives & Side Effects

Topical, the realistic reader concern

Redness and itching at the application site. Smith and Jacobson's 43-patient trial had two participants in each active group with temporary rash and itching, resolved by changing application. Kucera's 2018 series in 712 children recorded zero adverse events among the 386 treated on intact skin and one local burning and reddening reaction among the 326 treated on broken skin, a 0.14 percent intolerability rate. Koll 2004 reported no adverse drug reactions on active treatment.

Topical, the concern that is not resolved

Chronic low-level dermal alkaloid exposure has never been studied for carcinogenicity in humans. The 10-day labeled course exists because nobody has data past it. Melnyk and colleagues incubated comfrey root extract with skin microbiota from ten donors and found the alkaloid N-oxides deacetylated and deesterified but not converted to free alkaloids, which they read as suggesting intermittent external use is unlikely to pose substantial risk (Journal of Ethnopharmacology 2023, volume 318 Pt B, article 116968). An ex vivo finding, offered as a suggestion by its own authors.

Oral, the actual harm, and its denominator

Hepatic sinusoidal obstruction syndrome, formerly hepatic veno-occlusive disease: ascites, portal hypertension, right upper quadrant pain, weight gain from fluid retention, and in the severe form liver failure. At least one of the published comfrey cases ended in death, Yeong's 23-year-old man in 1990, and others required surgical hepatic and portal decompression.

At least six published English-language cases over four decades, the count turning on which reports you admit, is a small numerator against an unknown and probably large number of people who drank comfrey tea without apparent harm. That does not make the hazard small: the syndrome is under-recognized, latency can be years, and the dose that caused documented injury in the one quantified case, about 465 micrograms a day, is less than a 500 mg root dose delivers at Couet's contents.

Deficiency Symptoms

There is no comfrey deficiency: comfrey has no physiological role in humans and nothing in it is required for any human process.

What comfrey addresses

Nothing deficiency-related. Selenium has a requirement, a 55 microgram adult RDA, a deficiency disease in Keshan disease, selenoprotein P and glutathione peroxidase as biomarkers, and reversal on repletion. Comfrey has none of those five. Allantoin, its headline constituent, is a purine metabolite the body already makes.

Bottom line

Any product presenting comfrey as something a body runs short of has inverted the pharmacology.

Toxicity Symptoms

At high intake

The arithmetic below is a hazard calculation. None of these numbers is an amount to consume.

Work from Couet's commercial-sample measurements: 1,380 to 8,320 micrograms of total pyrrolizidine alkaloid per gram of root, 15 to 55 per gram of leaf. The reference point is 1.0 micrograms a day for a 50 kg adult, and the underlying acceptable intake of 0.0237 micrograms per kilogram per day, which the EMA itself scales to 0.5 micrograms a day for a 20 kg child.

The 450 mg leaf capsule that is actually on sale:

• 0.45 g x 15 to 55 = 7 to 25 micrograms, so 7 to 25 times the 1.0 microgram adult reference. A label reading two capsules twice daily multiplies it by four

• At the NTP document's upper dry-leaf figure of 2,000 micrograms per gram the same capsule carries 900

A 500 mg root dose, a worst-case hypothetical here rather than a product found on a US label:

• 0.5 g x 1,380 = 690 micrograms. 0.5 g x 8,320 = 4,160 micrograms, so 690 to 4,160 times the adult reference

• Staiger's compiled literature spans 0.013 to 1.2 percent, 130 to 12,000 micrograms per gram. At that floor a 500 mg dose carries 65 micrograms; at that ceiling, 6,000. Couet's commercial samples, used above, sit inside the span

• The root tinctures that are on sale state no extract ratio and no alkaloid figure, so neither calculation can be run on them

One cup of leaf tea, brewed on the EFSA protocol based on ISO 3103, 2.00 g in 150 mL of boiling water for 5 minutes:

• 2 g x 15 to 55 = 30 to 110 micrograms in the dry leaf

• Kaltner and colleagues measured transfer into a first 5-minute infusion from comfrey root at about 27 percent for the tertiary amines and about 50 percent for the N-oxides (Food Chemistry 2025, volume 489, article 145026). Comfrey alkaloid is almost entirely N-oxide, so use 50 percent

• A cup therefore delivers roughly 15 to 55 micrograms, 15 to 55 times the reference. Three cups a day is 45 to 165 times, and the UK FSA's 52.5 micrograms per gram retail infusion gives about 53 micrograms in one cup

• Root tea at the same brewing: 2,760 to 16,640 micrograms dry, roughly 1,380 to 8,320 in the cup

The one quantified human case, worked backward:

• Ridker and colleagues calculated that their 49-year-old patient consumed a minimum of 85 mg of pyrrolizidine alkaloid over 6 months, which they expressed as 15 micrograms per kilogram body weight per day (Gastroenterology 1985, volume 88, pages 1050 to 1054). The source was "a powder purporting to contain ground comfrey root"

• 15 micrograms per kilogram per day is 633 times the EMA's 0.0237 figure. The paper's own total implies about 465 micrograms a day, which puts the patient near 31 kg, not 70

• 465 micrograms a day is less than a 500 mg root dose delivers at Couet's measured contents, and roughly twenty to sixty times what the leaf capsule on sale delivers

• Bach, Thung and Schaffner's patient drank about 10 cups of comfrey tea daily plus comfrey pills for 8 years (American Journal of Medicine 1989, volume 87, pages 97 to 99). At 15 to 55 micrograms a cup that is 150 to 550 micrograms a day from tea alone, before the pills, across roughly 2,900 days

Two uncertainties cut against the low end of this. An assay that does not reduce the N-oxides before measurement undercounts by about an order of magnitude, the finding in the Public Health Nutrition analysis of commercial comfrey leaf teas. And Couet's leaf figure of 15 to 55 micrograms per gram is 36 to 130 times below the NTP document's upper dry-leaf figure of 2,000. Where two credible compilations disagree by two orders of magnitude on the same organ, treat the higher one as the planning number.

Signs to reduce or stop

• Topical: any redness, itching, burning or rash at the application site. Stop, do not reapply

• Oral, if someone has been taking it: abdominal distension or new ascites, right upper quadrant pain, unexplained weight gain from fluid retention, nausea, fatigue. This is an urgent medical evaluation, not a dose adjustment

General note

The hazard here is not a single large dose. Ridker's case established that low-level chronic exposure over months is sufficient, the authors' own conclusion. Comfrey toxicity accumulates, so "I have been fine on it for a year" describes exposure, not safety.

How Comfrey Works

Comfrey has two mechanisms: a local one in skin, probably useful and poorly characterized, and a hepatic one that only matters if you swallow it.

The hepatic route, in brief, because the arithmetic depends on it

Comfrey's alkaloids do not arrive as poisons. Intermedine, lycopsamine and symphytine reach the gut as water-soluble esters that are mostly excreted unchanged, and hepatic cytochrome P450 enzymes convert the 1,2-unsaturated fraction into a pyrrolic electrophile with an appetite for protein and DNA. That metabolite survives long enough to travel out of the hepatocyte and reach the sinusoidal lining, which is why the injury is vascular and why the regulatory numbers are single micrograms a day.

Allantoin and rosmarinic acid, the two named candidates

Staiger calls allantoin and rosmarinic acid "probably of central importance" to the pharmacodynamics, and that word is the state of the evidence: no study has removed either from a comfrey extract and shown the clinical effect disappear. Rosmarinic acid inhibits malondialdehyde formation in human platelets, prostaglandin synthesis, and carrageenan- and gelatin-induced erythrocyte aggregation, all in vitro.

The fractions nobody markets

A glycopeptide from comfrey root dose-dependently inhibited release of prostaglandin E2, prostaglandin I2, 12-HETE and arachidonic acid in rat stomach preparations. A 60 percent ethanolic root extract produced dose-dependent anticomplementary effects on complement activation. A 40 percent ethanolic extract and its fraction above 1,000 kD both inhibited shrinkage of a collagen matrix in a fibroblast model, the closest thing to a wound-healing mechanism in the file.

Mucilage, the mechanical contribution

The mucilage hydrates into a film, which plausibly explains the abrasion results without any pharmacology, and explains nothing about knee osteoarthritis.

So the active is unidentified. The trials measured outcomes in people, the mechanism work measured fractions in dishes, and nobody connected the two.

Synergistic Supplements

No comfrey combination has been tested against comfrey alone in a human trial, so no synergy is established. One is worth naming because it is sold: 35 percent comfrey root extract with 1.2 percent methyl nicotinate, licensed in Germany, Luxembourg and Switzerland, with a 162-patient observational study behind it. A 379-patient three-arm trial did separate the ingredients: the combination beat methyl nicotinate alone by 27 percent on the primary movement-pain area under the curve (163 against 164, with 52 on placebo, p less than 0.0001), while methyl nicotinate alone beat placebo by 31 percent. That sharpens the warning rather than softening it, because much of the effect is the vasodilator. For an acute sprain, what has evidence alongside topical comfrey is compression, elevation and an oral analgesic, which is what Koll and Predel permitted as rescue medication.

Interactions & What NOT to Take

Do not stack other pyrrolizidine alkaloid sources

Coltsfoot, butterbur or borage seed oil that is not certified alkaloid-free, Echium and Petasites preparations. The 1.0 microgram a day reference is a total across all sources, not a per-product allowance.

If someone is taking it orally, the drugs that matter

Anything with its own hepatic toxicity: acetaminophen at or near maximum dose, methotrexate, isoniazid, amiodarone, valproate, azathioprine, and alcohol. Drugs that induce hepatic drug-metabolizing enzymes, including rifampin, carbamazepine, phenytoin and St John's wort, would be expected to increase conversion of these alkaloids to their reactive form. That is mechanistic inference, with no comfrey case report demonstrating it.

Topically

Do not layer a topical NSAID on the same site. Predel 2005 compared comfrey ointment against diclofenac gel rather than combining them, and no trial has tested the combination.

Quality, Testing & Adulteration

The documented substitution, with its chemical marker

Echimidine and symlandine are absent from S. officinale and present in S. asperum and S. x uplandicum, so finding them in material labeled S. officinale indicates substitution. The EMA assessment report says their presence "may refer to falsification with Symphytum peregrinum roots," Symphytum peregrinum being a synonym of the hybrid. This is the single most useful assay in the comfrey file and no retail label reports it.

Documented misidentification, worse than adulteration here

The Digitalis purpurea incidents came from leaves gathered out of bloom. Wild-harvested leaf is the risk; a licensed cultivated product is not.

The specification problem

Three published ceilings for the most-tested extract: 0.35 and 35 ppm, a hundredfold apart, with 1 ppm for non-clinical batches. All are maxima, and none states which alkaloids it sums. Traumaplant's figure is a detection limit below 0.01 micrograms per gram, also not a measured content. Not one comfrey trial reports what its ointment actually contained.

The assay-scope trap

A certificate of analysis giving "total pyrrolizidine alkaloids" without stating whether N-oxides were reduced, which analytes were summed, and the limit of quantification is not interpretable, and a result reported as "less than" the limit is not a measurement of zero.

What third-party testing covers

I could not find a USP, NSF or ConsumerLab program that verifies a comfrey product or reports alkaloid content on one. Trifan and colleagues profiled 16 commercial root batches from 12 European countries by liquid chromatography with high-resolution tandem mass spectrometry in 2020, annotating 20 alkaloids and 17 phenolics, and proposed globoidnan A, globoidnan B and rabdosiin alongside rosmarinic acid as markers. That is a research method, not a commercial program.

What is checkable on the carton

Species on the label, S. officinale or S. x uplandicum or nothing. Plant part, root or aerial parts or leaf. Extract ratio and solvent. Percentage in the finished product. A named cultivar. A stated alkaloid limit. And the AHPA trade requirement statement, adopted July 1996 and revised July 2010: "For external use only. Do not apply to broken or abraded skin. Do not use when nursing." That binds AHPA members, not every seller, which is why oral comfrey is still on sale.

Special Considerations

Children

The EU monograph does not recommend root preparations under 18, though German labels for the same ointment cover age 3 and over and a 306-child series ran in ages 3 to 12. The herb cream's pediatric studies set a floor of 4: a 108-patient randomized trial, and a 712-child series whose protocol specified 4 to 12 years and logged its three 3-year-olds as deviations.

Gardeners and livestock keepers

Wear gloves, keep leaves out of the kitchen, and brew nothing from a plant you identified yourself.

Pregnancy and lactation

Both are contraindications, for the reasons given under Contraindications.

Research Status & Evidence Quality

Moderate evidence: topical pain reduction in acute ankle sprain, with two named preparations

Two placebo-controlled trials (Koll 2004, n = 142; Kucera 2004, n = 203) and one against diclofenac gel (Predel 2005, n = 164). Consistent direction, adequate size, sponsor-affiliated authorship.

Low to moderate evidence: acute back pain

Giannetti 2010 (n = 120) and Kucera 2005 (n = 215). The 2014 Cochrane review graded the Giannetti evidence low quality, an independent appraisal and the one to weight.

Low evidence: knee osteoarthritis and fresh abrasions

Grube 2007 (n = 220 full analysis set, 186 valid cases) and Smith and Jacobson 2011 (n = 43 against a eucalyptus cream that was not inert) for the knee. Barna 2007 (n = 278) and a pediatric trial (n = 108) for abrasions, both against a 1 percent cream: large effects from a dose comparison, not a placebo one.

Strong evidence: oral hepatotoxicity and animal hepatocarcinogenicity

At least six published English-language cases of sinusoidal obstruction syndrome, one quantified at 15 micrograms per kilogram body weight per day over 6 months, at least one fatal, and adenomas in every comfrey-fed group in Hirono 1978. LiverTox assigns a likelihood score of C for oral use.

No evidence: any oral therapeutic use

No controlled trial of swallowed comfrey exists for ulcer, colitis, arthritis or fracture healing.

Research Limitations

Every root-ointment trial carries Merck Selbstmedikation authorship, and both widely cited reviews were written by a Merck employee who discloses it. Harras Pharma Curarina sponsored the herb-cream work. Frost, Macpherson and O'Meara's 2013 scoping review in Complementary Therapies in Medicine (volume 21, pages 724 to 745) found that most included trials carried "an overall unclear risk of bias due to poor quality of reporting," and that the only multi-indication reviews then available were "written by an employee of a comfrey product manufacturer." No trial compared the two product families head to head, none reports a measured alkaloid content, and none measured a human plasma concentration. Single courses have been studied out to 12 weeks in a comfrey combination cream and to 200 days in one trial Frost included, but nobody has followed a repeat-course user across a year.

Summary & Key Takeaways

Bottom Line

Topical comfrey has real randomized evidence for acute sprain and acute back pain, and it belongs to two products: a 35 percent extract of S. officinale root in 60 percent ethanol, and a 10 percent preparation of the aerial parts of the S. x uplandicum cultivar 'Harras' with no detectable alkaloid. A capsule, a tea, a leaf powder or a cream labeled only "comfrey extract" has none of it. The 450 mg leaf capsule on US shelves carries roughly 7 to 25 micrograms of pyrrolizidine alkaloid against a reference of 1.0 micrograms a day, a cup of leaf tea 15 to 55, and a root tincture declares no figure at all.

Key Safety Points

• Do not swallow comfrey. The FDA asked firms to pull it on 6 July 2001 and never wrote the rule. That is a gap in enforcement, not a verdict on safety

• On intact skin, up to ten days with a labeled product, never on broken or irritated skin for a root preparation

• A homemade leaf poultice can put 105 to 385 micrograms against skin, 25 to 92 times a labeled ointment's load

• New abdominal swelling, right upper quadrant pain or unexplained weight gain in anyone with an oral comfrey history is an urgent evaluation, and a normal liver panel does not rule it out

Special Note

Comfrey is this archive's clearest case of evidence borrowed across a species line. Hirono's carcinogenicity paper is titled Symphytum officinale and describes Russian comfrey; the abrasion trials used the hybrid, the sprain and osteoarthritis trials the true species' root. When a label says only "comfrey," nobody has told you which plant, which part, or how much alkaloid.

Researched and drafted with AI assistance. Every claim verified against primary sources and human-reviewed before publication. Last reviewed: [DATE]. Corrections: reply to any issue.

This issue is educational and is not medical advice. Comfrey is a plant with a withdrawn oral use and a narrow licensed topical one, and nothing here is a recommendation to take it. If you are on prescription medication, have a diagnosed liver condition, are pregnant or breastfeeding, or are treating an injury in a child, talk to a clinician before starting or stopping anything.