Boneset: Immune and Preclinical Research

Boneset has a small, genuine, and reasonably careful modern research literature. Most of it comes from a handful of groups, much of it is chemistry rather than pharmacology, and essentially all of the pharmacology is in vitro. This page sets out what that literature found, in the order the evidence deserves rather than the order that flatters the plant — which means the absence of human trials comes first, not last, and the famous influenza finding is placed in the middle of the page rather than at the top of it.

The reason for that ordering: on boneset, the single best-established fact is a safety fact — the plant belongs to a group of species in which dehydro-pyrrolizidine alkaloids have been analytically detected, and that alkaloid class has a characterised human hepatotoxicity. The benefit claims are all one or two tiers below that. Presenting the cell-culture antiviral result as the headline and the alkaloids as a footnote would invert the actual strength of the evidence.

Where the Evidence Actually Stands
What Is Actually in the Plant
The Immunostimulant Polysaccharide Work
The Influenza Attachment Finding
Anti-Inflammatory Activity in Cell Models
Antiprotozoal, Antibacterial and Cytotoxicity Screens
Antioxidant Assays and Why They Prove Little
Borrowed Evidence: Four Kinds Found Here
From a Dish to a Cup: Doing the Arithmetic
What Is Missing, Stated as Findings
What an Adequate Trial Would Look Like
The Honest Verdict
Key Research Papers
Connections

Where the Evidence Actually Stands

Ranked from strongest to weakest, boneset's evidence base looks like this. Note that the top three entries are all about chemistry or safety, and none is about benefit.

Established (analytical chemistry). The plant's constituent profile has been characterised by identifiable groups: sesquiterpene lactones including guaianolide and germacranolide types, flavonoids, caffeic acid derivatives, and polysaccharides. This is solid work and is not in dispute.
Established (class-level human toxicology). Dehydro-pyrrolizidine alkaloids cause hepatic sinusoidal obstruction syndrome in humans. That is not a boneset finding — it is a finding about the compound class — but analytical detection of such alkaloids in E. perfoliatum connects the plant to it.
Reproducible in vitro (immunological and anti-inflammatory). Extracts and fractions have shown immune-cell activation and inhibition of inflammatory signalling in cell models, across more than one publication.
Single-study in vitro (antiviral). A hydroalcoholic extract has been reported to interfere with influenza A virus attachment in cell culture.
Absent (human clinical). No adequate controlled human trial of the herb for any indication.
Absent (human pharmacokinetics). No published absorption, distribution or plasma-concentration data for the plant's characteristic constituents from any human preparation.
Absent (dose-finding). No dose established against a measured clinical outcome.

That is the whole picture. Everything below is detail on the middle tiers.

What Is Actually in the Plant

Boneset's chemistry is the best-supported thing about it, and it is genuinely interesting. Four groups matter.

Sesquiterpene lactones. The plant's signature constituents and the source of its extreme bitterness. The foundational isolation work in the 1970s named compounds after the plant itself — euperfolin, euperfolitin, eufoliatin and relatives — and later work by the Münster group added further structures, including an unusual dimeric guaianolide. Sesquiterpene lactones as a class are electrophilic Michael acceptors: they react with thiol groups, which is the general mechanism behind both their anti-inflammatory activity (they can inhibit NF-κB signalling) and their cytotoxicity and contact-allergenic potential. Benefit and hazard here are one chemical property described twice.
Flavonoids. Including eupafolin (also known as nepetin, a 6-methoxylated flavone), eupatorin, and quercetin- and kaempferol-type derivatives. Important caveat below: these compounds are not exclusive to boneset.
Caffeic acid derivatives. Chlorogenic-acid-type phenolics, characterised specifically in this plant. These are the constituents most likely to be substantially present in a water infusion, being polar and water-soluble.
Polysaccharides. High-molecular-weight sugar polymers, the fraction most often invoked in immunostimulant explanations for cold-and-flu herbs.
Dehydro-pyrrolizidine alkaloids. Detected analytically in E. perfoliatum and related species. Not a benefit constituent, and the reason this set of pages carries a dedicated safety article.

Notice what this list implies about preparation. A hot water infusion will extract the polar phenolics and much of the polysaccharide fraction efficiently; it will extract sesquiterpene lactones and flavonoid aglycones less efficiently; an ethanolic tincture reverses that bias. A tea and a tincture are not the same intervention, and a result obtained with one cannot be transferred to the other without argument. This is not a hypothetical concern on this plant — the antiviral work used a hydroalcoholic extract, and the traditional preparation is a tea.

The Immunostimulant Polysaccharide Work

The oldest immunological finding on boneset dates to the mid-1980s, when Wagner's group in Munich reported immunologically active polysaccharides from Eupatorium species. Polysaccharide fractions of this kind typically show activity in macrophage or granulocyte assays — phagocytosis stimulation, or activation readouts in cell-based systems.

Three things must be said about it.

First, the study was of two species: Eupatorium cannabinum (hemp agrimony, a European plant) and Eupatorium perfoliatum. Where results are reported for the pair, they are not results for boneset. This is species substitution operating inside a single paper's title, and it is easy to miss.

Second, immunostimulant activity in a polysaccharide fraction is close to a generic finding in plant pharmacology. Comparable results have been reported for echinacea, astragalus, mushroom beta-glucans, aloe, and dozens of others. That does not make the finding false; it makes it non-specific. It tells you the plant contains polysaccharides that do what plant polysaccharides do in these assays.

Third and most limiting: large polysaccharides are not meaningfully absorbed from the gut. A molecule of that size does not cross the intestinal epithelium intact into the circulation. Any oral effect would have to be mediated locally — gut-associated lymphoid tissue, pattern-recognition receptors on intestinal immune cells, or fermentation to short-chain fatty acids by the microbiota. Those are real possible routes and they are actively researched for other plant polysaccharides. They are also entirely different mechanisms from the one implied when a supplement label says a herb “activates immune cells,” and none of them has been demonstrated for boneset.

The Influenza Attachment Finding

This is boneset's most-cited modern result and the one most likely to appear in marketing, so it deserves precision. Work published in 2016 in the Journal of Ethnopharmacology by Derksen, Kühn, Hafezi and colleagues, from the same Münster group that produced most of the plant's chemistry, reported that a hydroalcoholic extract of E. perfoliatum interfered with the attachment of influenza A virus to host cells in culture.

What makes it interesting: the mechanism is specific rather than vague. Interference at the attachment step is a defined stage of the viral life cycle, it is measurable, and it lines up with the plant's traditional indication in a way that is not obviously coincidental. This is the kind of result that justifies further work, and this site treats it as a legitimate lead.

What it does not establish:

It is cell culture. Extract and virus were together in a controlled system at a controlled concentration. Nothing about absorption, distribution, metabolism, protein binding or delivery to respiratory epithelium is addressed.
Attachment interference is a topical-contact mechanism. If the effect requires extract in physical contact with virus and cell surface, then an oral infusion — digested, absorbed if at all, and diluted into body water — is a poor delivery route for it. The mechanism arguably argues against the traditional preparation rather than for it.
The extract was hydroalcoholic. The tradition is a hot water infusion. Different solvent, different constituent profile, different intervention.
No animal challenge study followed it into an infected organism, and no human study followed that.

The honest formulation: a boneset extract has a plausible, specific, in-vitro anti-influenza mechanism, and there is no evidence that drinking boneset tea does anything to influenza in a person. Both halves of that sentence are load-bearing.

Anti-Inflammatory Activity in Cell Models

The best-replicated pharmacological signal on this plant is anti-inflammatory. Work from the same group reported that E. perfoliatum extracts, the flavone eupafolin, and a dimeric guaianolide inhibited inducible nitric oxide synthase activity and modulated inflammation-related cytokines and chemokines in cell models.

Mechanistically this is coherent. Sesquiterpene lactones inhibit NF-κB, which sits upstream of iNOS and of most of the cytokines measured; methoxylated flavones have well-described activity in the same pathways. If any part of boneset's traditional reputation has a plausible pharmacological basis, this is it — not curing the infection, but blunting the inflammatory component of the illness that produces the aching and the malaise. That is precisely the traditional indication as the old texts describe it, which is a satisfying convergence.

It is also where the compound substitution problem bites hardest. Eupafolin is not a boneset compound; it is a flavone that occurs across many plants and has its own substantial pharmacology literature, most of it conducted with material from other sources. A reader who searches eupafolin will find a great deal of anti-inflammatory and anticancer cell work, and almost none of it is about boneset. Crediting that literature to boneset would be the same error as crediting rosemary's rosmarinic acid pharmacology to perilla, or green tea's catechin literature to jasmine tea. The relevant question is not “what does eupafolin do?” but “how much eupafolin does a cup of boneset tea deliver, and to where?” — and that question has no published answer.

Antiprotozoal, Antibacterial and Cytotoxicity Screens

Two further in-vitro strands exist, and both are more equivocal than they first appear.

The chemistry paper describing the unusual dimeric guaianolide also reported antiprotozoal activity for isolated constituents, tested against the parasite panels standard in that field. This is a routine screening exercise attached to a structure-elucidation paper rather than a therapeutic programme, and sesquiterpene lactones are frequently active in such screens because of their general thiol reactivity. It should not be read as a suggestion that boneset treats a parasitic infection.

Separately, a 2000 paper in Phytotherapy Research by Habtemariam and Macpherson tested an ethanol extract of boneset leaves and reported both antibacterial activity and cytotoxicity. The second half of that title is the more informative half. Cytotoxicity in a plant extract containing electrophilic sesquiterpene lactones is expected, and it is the same chemistry as the anti-inflammatory activity. A compound class that reacts with cellular thiols does not distinguish between an inflammatory signalling protein and any other thiol it meets. When a screen finds that an extract kills bacteria and harms mammalian cells, the selectivity ratio is the whole story, and general antimicrobial activity in a crude extract is one of the least informative findings in phytochemistry.

Antioxidant Assays and Why They Prove Little

Boneset extracts scavenge free radicals in the standard chemical assays. So does every plant with a meaningful phenolic content, which is nearly all of them. Coffee, tea, red onion skin, and the herbs in a supermarket spice rack all perform well in these tests.

The assays measure the chemical reducing capacity of a solution in a tube. They do not measure antioxidant activity in a human body, where a compound must first survive digestion, be absorbed, escape first-pass metabolism, reach a tissue at a relevant concentration, and act there in competition with endogenous systems that are far more abundant than any dietary phenolic. The gap between the two is the reason antioxidant capacity figures have largely been abandoned as a meaningful nutritional claim. On this page they are noted for completeness and are not treated as supporting any benefit.

Borrowed Evidence: Four Kinds Found Here

Boneset attracts an unusual number of substitution errors for such a small literature. Each is flagged at the point of use elsewhere on these pages; collected here, they are:

Species substitution. Work on Eupatorium cannabinum, on E. odoratum and relatives, or on plants now placed in Eutrochium and Ageratina, cited as boneset data. The genus has been dismembered taxonomically, so an older paper's genus name may not point where a modern reader assumes. A citation reading “Eupatorium” is not automatically a citation about E. perfoliatum.
Preparation substitution. Homeopathic Eupatorium perfoliatum studies cited as evidence for the herb. A serial dilution of a mother tincture is a different intervention in a different framework, by orders of magnitude in concentration. This is the largest single source of apparent “human evidence” for boneset and it is not human evidence for boneset.
Compound substitution. The eupafolin literature, and general sesquiterpene lactone or flavonoid pharmacology, credited to the plant without any account of how much of the compound a real preparation delivers.
Solvent substitution. Hydroalcoholic-extract results — including the influenza attachment finding — cited to support a hot water infusion. Different solvents extract different constituents in different proportions; this is basic and it is routinely elided.

Strip all four out and what remains is a small, honest, mostly in-vitro literature by a few groups, plus a large historical record. That is the actual state of knowledge.

From a Dish to a Cup: Doing the Arithmetic

Hedging is less persuasive than arithmetic, so here is a worked illustration of the gap between an in-vitro concentration and an oral dose. The input numbers below are illustrative, chosen to show the shape of the calculation — they are not any specific study's figures, and no specific study's figures are asserted here. The point is the order of magnitude, and it does not change much across plausible inputs.

Assume a crude extract shows an effect in cell culture at a concentration of the order of 100 micrograms per millilitre — an entirely typical figure for crude plant extract screens, and often on the optimistic side. Now ask what it would take to reach that concentration in a person:

Total body water in a 70 kg adult is roughly 42 litres, or 42,000 mL.
100 µg/mL across 42,000 mL is 4,200,000 µg, which is about 4.2 grams of extract.
That assumes 100% oral bioavailability, no first-pass metabolism, no protein binding, no glucuronidation of the phenolics, and even distribution — every one of which is generous to the herb, several of them wildly so. Polyphenols are typically absorbed at single-digit percentages and conjugated rapidly; large polysaccharides are essentially not absorbed at all.
A cup of tea made from a couple of grams of dried herb yields a fraction of a gram of total extractable solids, of which the compound of interest is a small percentage. The active fraction delivered is plausibly in the milligram or sub-milligram range.

The shortfall is therefore not a factor of two or three. It is three or four orders of magnitude before any correction for bioavailability, and larger after. This is the same arithmetic that reduces most in-vitro essential-oil antimicrobial claims to implausibility, and it applies with full force here.

Two honest counterweights, because the calculation is not the end of the argument. First, if a mechanism acts locally — in the gut lumen, on gut-associated immune tissue, or on the mucosa of the mouth and throat as the tea passes — then systemic dilution is the wrong denominator and the concentrations achieved locally can be far higher. That is a real caveat and it is why the polysaccharide and mucosal stories are not dismissible. Second, the calculation says nothing about whether a low-concentration effect exists; it says the specific in-vitro concentration is not reachable systemically. What it rules out is the casual inference from “active in a dish” to “active in a person” — which is exactly the inference this literature is most often used to make.

What Is Missing, Stated as Findings

The following are results of this review, not gaps in it. Each is a stated conclusion that the relevant work has not been done.

No randomised controlled trial of boneset herb exists for any indication. Not for influenza, not for febrile illness, not for musculoskeletal pain, not for digestion.
No human pharmacokinetic data exist for any characteristic boneset constituent from any preparation. Peak plasma concentration, half-life and bioavailability are all unknown, which makes in-vitro-to-human extrapolation impossible in principle rather than merely difficult.
No animal infection-challenge study has followed the in-vitro antiviral result into an infected organism.
No formal dose-finding study exists. Traditional dosing was bounded above by emesis, not established against an outcome.
No systematic quantification of pyrrolizidine alkaloid content across commercial products is publicly available. Analytical detection in plant material is documented; a survey of what is actually in retail boneset on a per-batch basis is not.
No drug interaction study of any kind. “No known interactions” for boneset means no interaction study has been performed. That is absent data, not reassurance — and it is worth noting that the plant's constituent classes include compounds that plausibly touch hepatic metabolism.
No modern comparative identification survey of retail material, despite the genus having been taxonomically split and despite a toxic look-alike growing in the same habitat.
No standardised extract with a defined marker compound is in general use, so even if a trial were run it would be hard to say what had been tested.

Compare this with the state of evidence for herbs occupying the same shelf. Elderberry preparations, Pelargonium sidoides, Andrographis paniculata and echinacea preparations all have human randomised trial data of varying and often criticised quality — but data that exists, can be argued over, pooled, and revised. Boneset's list above is what “traditional herb with no modern research programme” actually looks like, and naming the comparators pre-empts the objection that nobody ever trials traditional plants.

What an Adequate Trial Would Look Like

It is worth being concrete about what would change the picture, because it clarifies how far away that is.

A characterised preparation. A defined extract of authenticated E. perfoliatum, with a marker compound quantified batch to batch, and with pyrrolizidine alkaloid content measured and reported — because an intervention with unquantified hepatotoxin content is not ethically deliverable to volunteers in the first place.
Human pharmacokinetics first. A small study establishing whether anything reaches the circulation, at what concentration, and for how long. Without this step the trial cannot be designed rationally.
A defined population and outcome. Laboratory-confirmed influenza within a specified window of symptom onset, with pre-registered primary outcomes — time to symptom resolution on a validated scale, or fever duration — rather than a global impression.
Placebo control and blinding, which is hard here. Boneset is ferociously bitter and can induce nausea; a credible placebo has to match that, or unblinding is immediate and the trial measures expectation.
Liver monitoring as a safety endpoint, given the alkaloid class involved.
Adequate size. Symptom-duration outcomes in self-limiting respiratory illness need hundreds of participants to detect the modest effects that are realistically on offer.

None of this is exotic; it is the ordinary requirement. The realistic assessment is that it is unlikely to happen. Boneset is off patent, commercially minor, and carries an alkaloid liability that makes ethics review harder than for a comparable non-PA herb. The evidence gap on this plant is probably permanent, and saying so is more useful to a reader than implying that clarity is coming.

The Honest Verdict

Boneset is chemically interesting, historically important, and clinically unevaluated. Its anti-inflammatory in-vitro signal is coherent with what the plant was traditionally used for, which is more than can be said for many folk remedies. Its influenza attachment finding is a real and specific lead. And the total of human evidence for any benefit is zero, while the plant belongs to a group in which a compound class with established human hepatotoxicity has been analytically detected.

That asymmetry is the finding. It is not a reason to sneer at the plant — it is a genuine piece of North American medical history and a genuinely good research subject. It is a reason to treat any product on a shelf claiming boneset benefits as making a claim with nothing behind it, and to read the safety article before deciding anything.

Key Research Papers

Given as PubMed topic searches so links keep working and so a reader sees a question's whole literature. Where a paper's metadata could not be confirmed here, the finding is described and a topic search is given instead of a specific attribution.

Hensel A, Maas M, Sendker J, and colleagues. “Eupatorium perfoliatum L.: phytochemistry, traditional use and current applications.” Journal of Ethnopharmacology, 2011. Start here; it is the review that frames everything else on this page. Find on PubMed.
Derksen A, Kühn J, Hafezi W, and colleagues. “Antiviral activity of hydroalcoholic extract from Eupatorium perfoliatum L. against the attachment of influenza A virus.” Journal of Ethnopharmacology, 2016. The attachment-interference result. In vitro; hydroalcoholic extract, not an infusion. Find on PubMed.
Maas M, Deters AM, Hensel A. Anti-inflammatory activity of Eupatorium perfoliatum extracts, eupafolin and a dimeric guaianolide via inducible nitric oxide synthase inhibition and modulation of inflammation-related cytokines and chemokines. Journal of Ethnopharmacology, 2011. The best-replicated pharmacological signal on the plant. Find on PubMed.
Maas M, Hensel A, Schmidt TJ, and colleagues. An unusual dimeric guaianolide with antiprotozoal activity and further sesquiterpene lactones from Eupatorium perfoliatum. Phytochemistry, 2011. Structure elucidation with an attached parasite screen. Find on PubMed.
Vollmar A, Schäfer W, Wagner H. Immunologically active polysaccharides of Eupatorium cannabinum and Eupatorium perfoliatum. Phytochemistry, 1986. Species substitution warning: a two-species study, frequently cited as though it were boneset-only. Find on PubMed.
Maas M, Petereit F, Hensel A. Caffeic acid derivatives from Eupatorium perfoliatum. Molecules, 2009. The water-soluble phenolic fraction — what a tea most plausibly delivers. Find on PubMed.
Habtemariam S, Macpherson AM. Cytotoxicity and antibacterial activity of an ethanol extract from leaves of the herbal drug boneset. Phytotherapy Research, 2000. Read the cytotoxicity half as carefully as the antibacterial half. Find on PubMed.
Herz W and colleagues. Sesquiterpene lactones of Eupatorium perfoliatum. Journal of Organic Chemistry, 1977. The foundational isolation of euperfolin and euperfolitin. Find on PubMed.
Sesquiterpene lactones as NF-κB inhibitors and thiol-reactive electrophiles. The mechanistic literature explaining why one chemical property underlies both the anti-inflammatory activity and the cytotoxicity and contact allergy. Search PubMed on this topic.
Eupafolin (nepetin) pharmacology. The literature on this flavone, most of it conducted with material from plants other than boneset — the clearest example of compound substitution on this page. Search PubMed on this topic.
Oral bioavailability of dietary polyphenols and plant polysaccharides. Why the arithmetic above comes out the way it does, and where the local-action caveat comes from. Search PubMed on this topic.
Herbal interventions for acute respiratory infection that do have randomised human data — elderberry, Pelargonium sidoides, Andrographis, echinacea. The comparator set. Search PubMed on this topic.
Dehydropyrrolizidine alkaloids in Eupatorium perfoliatum and related species. Analytical detection work by Betz, Colegate, Upton and colleagues — the highest-tier fact on this plant. Search PubMed on this topic.

Back to Table of Contents

Connections

Back to Table of Contents

A note on what this page is

This is a review of the published preclinical literature, written for a general reader. It is not medical advice and it is not a recommendation to use boneset. The plant contains pyrrolizidine alkaloids; it should not be used in pregnancy, while breastfeeding, in infants or children, or by anyone with liver disease, and it is not suitable for long-term or repeated use. If you are unwell or taking medication, talk to a qualified healthcare professional.