Houttuynia Safety: Family Questions, Dose and Unknowns

A reader who has heard of aristolochic acid and then reads that Houttuynia cordata sits in the same botanical order as the birthworts is entitled to be worried, and to be given a straight answer rather than either a dismissal or a scare. Aristolochic acid is one of the genuinely dangerous compounds in the herbal world: a potent nephrotoxin and a human carcinogen, responsible for an epidemic of kidney failure and urothelial cancer. If fish mint carried it, that would be the only thing worth saying about the plant.

It does not. Houttuynia cordata is not an established aristolochic-acid-containing plant, analytical testing of correctly identified material has not found aristolochic acids in it, and it should not be presented as if it did. Getting that wrong in the alarmist direction is as much an error as getting it wrong in the reassuring one — manufacturing a hazard is not caution.

What survives the question is narrower and more interesting than the question itself, and it is worth the page: what order-level relatedness is and is not worth as evidence; why the compounds Houttuynia actually contains are chemically downstream of the dangerous ones rather than equivalent to them; why the real historical hazard in this corner of herbal medicine has been substitution rather than the correct species; and why a certificate of analysis showing a marker compound tells you nothing about which plant is in the jar.

Table of Contents

1. Piperales: What Order-Level Relatedness Is Worth
2. Aristolochic Acids and Aristolactams Are Not the Same Compound
3. What Analytical Testing Found
4. The Real Historical Hazard Was Substitution
5. A Marker Assay Is a Potency Check, Not an Identity Check
6. What Remains Genuinely Open
7. Other Safety Points, Handled Briefly
8. Hazards Houttuynia Does Not Have
9. Key Research Papers
10. Connections

Piperales: What Order-Level Relatedness Is Worth

The botanical fact is correct, and worth stating precisely. Houttuynia cordata belongs to the Saururaceae, the lizard's-tail family — a small group of a few genera. Saururaceae sits in the order Piperales. So does the Aristolochiaceae, the birthwort family, which is where aristolochic acid comes from. Same order, different families.

The clean way to see what that is worth is to look at the third major family in the same order: the Piperaceae. Black pepper is a Piperaceae. So is kava. Nobody has ever suggested testing black pepper for aristolochic acid, and nobody should — because taxonomic order is far too coarse a unit to predict secondary chemistry.

Some scale for that. An order groups families that shared an ancestor tens of millions of years ago. Within one order you routinely find families whose useful chemistry has nothing in common. The relevant secondary-metabolite pathways are typically distributed at the level of family, genus, or even a subset of species within a genus — which is why Artemisia annua makes artemisinin and its close relatives largely do not.

The same reasoning applies inside the Aristolochiaceae in the direction of caution, and that is the more useful half:

• Aristolochia species — birthworts, including A. fangchi, A. manshuriensis and A. clematitis — are the classic sources, and are the plants at the centre of the poisoning episodes.
• Asarum species — wild ginger, the Chinese herb xi xin — are in the same family and are a documented aristolochic-acid concern, which is why regulators in several countries restrict them.
• Saururaceae — Houttuynia's family — is not an aristolochic-acid family. The compound class is not characteristic of it.

So the honest summary of the taxonomy point: the family is the informative unit here, and the family is clean. Order membership prompted a reasonable question, the question was answered analytically, and the answer did not depend on the taxonomy at all. That is how it should work — relatedness generates a hypothesis, chemistry tests it.

Aristolochic Acids and Aristolactams Are Not the Same Compound

Houttuynia does contain compounds in a related structural class: aristolactams, principally aristolactam I and aristolactam BII. This is the true premise underneath the worry, and it needs to be handled chemically rather than by name similarity — the names are confusingly close in English and closer still in Chinese, which is part of why the anxiety spread.

The difference that matters is the mechanism of harm:

1. Aristolochic acids carry a nitro group. They are nitrophenanthrene carboxylic acids, and their toxicity depends on that nitro group being reduced in the body. Nitroreduction generates a reactive intermediate — a nitrenium ion — which binds covalently to DNA, forming adducts on adenine. Those adducts produce the characteristic A-to-T mutation signature found in the tumours of exposed people, including in the TP53 gene.
2. Aristolactams do not carry that nitro group. Chemically they sit downstream: aristolactam I is a known metabolite of aristolochic acid I, formed after the reactive step has happened. It is an end product of the activation pathway, not a precursor that can enter it.

That distinction is the heart of the matter. A compound whose danger comes from being metabolically activated is a different proposition from a compound that is already the deactivated end product. It does not make aristolactams inert — aristolactam I has been reported as toxic to isolated human kidney cells in culture, and in one comparison more so than aristolochic acid I itself — but cell-culture toxicity is not what happens in a body, which is the next section's point in miniature.

A 2025 study in Food Research International fed aristolactam I, aristolactam BII and a whole aristolactam fraction from Houttuynia to mice, using aristolochic acid I as a positive control. The positive control did what it always does — clear kidney injury. The aristolactams produced only modest changes in body weight, serum kidney markers and urinary injury biomarkers. The explanation offered is pharmacokinetic rather than a claim that the compounds are harmless: oral bioavailability of aristolactam I was reported at roughly half, against essentially complete absorption for aristolochic acid I; it was metabolised into less toxic derivatives; and no unchanged aristolactam appeared in urine. Poor absorption and rapid metabolism kept it away from the kidney.

Two caveats belong with that, because a single mouse study is a single mouse study. It is one experiment, in one species, at whatever doses and duration it used — this page does not quote a milligram-per-kilogram figure it cannot verify — and rodent pharmacokinetics do not automatically transfer to humans. The finding is genuinely reassuring in direction. It is not a human safety demonstration, and it is the only animal study of its kind on these specific compounds from this plant.

What Analytical Testing Found

The direct question — does the plant make aristolochic acids? — has been answered directly, which is the best possible outcome for a question of this kind.

A Hong Kong group used liquid chromatography with tandem mass spectrometry, the reference technique for this analysis, on twenty samples of fresh and sun-dried Houttuynia cordata collected from different Chinese cities, with detection limits in the low nanograms-per-gram range. Aristolochic acids I and II were absent or below the limit of detection in every sample. The authors' conclusion was that eating Houttuynia is unlikely to cause aristolochic acid nephropathy, because the plant is not making the carcinogens. The work was published in the Journal of Mass Spectrometry.

Three things about that study are worth a reader's attention, and one is a limitation:

• It measured the right thing. Not a proxy, not a related compound, not a computational prediction — the actual carcinogens, by the method regulators use.
• It covered both forms and multiple sources. Fresh and sun-dried, from several cities, which addresses the obvious objection that content might vary by region or processing.
• Twenty samples is a modest number. It is enough to establish that aristolochic acid production is not a normal feature of this species; it is not enough to guarantee that no individual batch anywhere is contaminated — and contamination, as the next section explains, is a supply-chain question rather than a botanical one.

The Real Historical Hazard Was Substitution

Every documented episode of aristolochic acid poisoning through herbal medicine involved a plant that was not what the label said, or a plant whose aristolochic acid content was never in question. Reading those episodes as "Chinese herbs cause kidney failure" gets the causal structure wrong, and reading them as "it was only an adulteration, so the herb is fine" gets it wrong in the other direction.

The defining case is Belgian. In the early 1990s a cluster of young women developed rapidly progressive interstitial renal fibrosis after attending a weight-loss clinic; the case series appeared in The Lancet in 1993. The preparation contained a Chinese herb, and the herb supplied was Aristolochia fangchi rather than the intended Stephania tetrandra. The two share a similar Chinese name — fang ji — and were confused in the supply chain. Many of the affected women went on to end-stage renal failure, and a follow-up study reported in the New England Journal of Medicine in 2000 found urothelial carcinoma in a substantial proportion of those who had their urinary tracts removed. Separately, work published in PNAS in 2007 linked aristolochic acid to Balkan endemic nephropathy, where the exposure route appears to have been Aristolochia clematitis seed contaminating wheat, and a 2012 PNAS study connected aristolochic acid exposure to the very high rate of upper urinary tract cancer in Taiwan. The International Agency for Research on Cancer classifies aristolochic acid mixtures and plants containing them as carcinogenic to humans.

Now the part that must not be skipped. Attributing that harm to substitution clears the plant, not the bottle:

1. The substitution happened because names collide. Fang ji in Belgium; more broadly, similar herb names, similar dried appearances and long informal supply chains. Those conditions have not disappeared.
2. If you cannot tell which plant you bought, the hazard is worse, not better. "It was a different species" is only reassuring if you can verify which species you have. For a dried, powdered, encapsulated product, most buyers cannot.
3. Most implicated products were never analysed. The reattribution to Aristolochia rests on the batches that were investigated. That is normal and it is also a limit on how much the reattribution proves.
4. The consequence points at product form, not at the species. Fresh fish mint bought as a vegetable, with roots and soil attached, in a Vietnamese grocery, is a plant you can see, smell and identify. An imported capsule labelled "Houttuynia extract" is a powder you cannot. The risk difference between those two purchases has nothing to do with botany.

A Marker Assay Is a Potency Check, Not an Identity Check

Supplement labels answer the verification problem with a certificate of analysis: "standardised to X% quercitrin," "assayed for total flavonoids." That is worth less than it looks, and the reason is worth understanding once, because it applies to every herbal product a reader will ever buy.

An assay for a marker compound measures how much of that compound is present. It does not establish which plant the compound came from. Quercitrin, rutin, hyperoside and chlorogenic acid — the compounds a Houttuynia product would plausibly be standardised to — are distributed across enormous swathes of the plant kingdom. A product could hit its quercitrin specification using material that is largely something else entirely, and the certificate would pass.

What does establish identity:

• DNA barcoding — sequencing standard marker regions such as ITS2, rbcL or psbA-trnH and matching them to reference sequences. This identifies the species, which is the actual question, though it degrades with heavily processed or extracted material where DNA is fragmented.
• Macroscopic and microscopic authentication of the raw herb by a trained analyst, the traditional pharmacognosy method, which works well on whole dried material and not at all on a powder.
• A full chromatographic fingerprint rather than a single marker — comparing the whole profile against an authenticated reference, which can detect a wrong species even without sequencing.
• Targeted testing for the specific hazard — for anything in this corner of herbal medicine, that means an LC-MS/MS screen for aristolochic acids, exactly as the Hong Kong study performed.

The practical conclusion is unglamorous and reliable: buy the vegetable, not the capsule. Fresh bunches sold as food in a Vietnamese or Chinese grocery, or dried dokudami-cha from a Japanese supermarket, are products whose identity you can verify with your own eyes and nose. That is a better authentication than most certificates of analysis, and it costs less.

What Remains Genuinely Open

Naming the unknowns as findings, rather than leaving them to be inferred:

1. No long-term human study of concentrated extract exists. Not a negative result — an absent one. Nobody has followed people taking a standardised Houttuynia extract daily for years with renal function and urinary-tract outcomes. That study has not failed; it has not been done.
2. No human pharmacokinetics for the aristolactams. The reassuring bioavailability argument comes from one mouse study. The corresponding human numbers do not exist.
3. No aristolactam content figure for the plant appears here. Published values would depend on plant part, origin, season, drying and analytical method, and this page could not verify a representative figure. A confident milligrams-per-gram number would be a guess, and it would be the number readers quoted afterwards.
4. No established safe or unsafe dose, for any form. There is no therapeutic dose because there is no proven therapeutic effect, and there is no toxic threshold because no dose-finding toxicology has been done in humans.
5. No data in pregnancy or lactation beyond customary practice. Culinary amounts have a long record of use including by pregnant women in Vietnam and Japan, with no signal of harm. That is an observation, not a study. For medicinal doses, extracts and capsules there is no safety data in pregnancy, and the sensible course is to avoid those forms.
6. No interaction studies. None, in either direction, with any prescription medicine. "No known interactions" for this plant means nobody has looked.
7. No paediatric safety work. Vietnamese and Japanese practice includes giving the juice and the tea to children, which is a real observation about tolerability at those amounts, and there is no formal study.

Who should act on this list rather than merely note it: people with chronic kidney disease or a history of acute kidney injury, anyone with a urothelial cancer history, transplant recipients, and anyone pregnant. For those readers the recommendation is specific and narrow: the food and the ordinary tea are not the issue; concentrated extract capsules are the form to discuss with a clinician before starting, and the form with nothing behind it.

Other Safety Points, Handled Briefly

• Allergy is plausible and rarely reported orally. Given the injectable hypersensitivity history, allergy to this plant is not a far-fetched idea — but the injection mechanism appears to be direct mast-cell activation requiring bloodstream delivery, which eating does not provide. See the injectable page for the full argument. Contact irritation in people handling large quantities of the fresh plant is the most plausible report. Stop if you develop rash, itching or swelling.
• Never use an injectable preparation. The one hard rule on these pages, repeated here because a safety page should not require a reader to visit another one to find it.
• Gastrointestinal upset from large raw servings. Some people get loose stools from a big plate of the raw herb. The traditional Chinese caution against long-term use in "spleen-and-stomach deficiency cold" maps onto exactly this observation, which is a case of a traditional framing tracking something real.
• Diuretic reputation and potassium. Traditionally described, never measured in a controlled human study, and the potassium implications are unmeasured in both possible directions. Covered on the food and tradition page; the group who should care is people on diuretics, ACE inhibitors, ARBs, potassium-sparing agents or lithium, and people with kidney disease.
• Do not eat ornamental nursery stock. The variegated "chameleon plant" sold as a groundcover is the same species, but garden-centre plants are grown as ornamentals and may carry pesticides not permitted on food crops. Buy it where it is sold as food.
• The smell is not a warning. The fishy aroma is medium-chain aldehydes. Odour intensity is a fact about volatility and olfactory receptor binding, not a toxicity signal.

Hazards Houttuynia Does Not Have

Readers arriving from other herb pages carry specific worries with them, and clearing the ones that do not apply is as much a part of an accurate safety page as listing the ones that do. On current evidence, Houttuynia cordata is not associated with:

• Aristolochic acid nephropathy. Analytical testing of correctly identified material has not found the compounds. This is the reassurance the page opened with and it is the best-supported item on the list.
• Pyrrolizidine alkaloids. The hepatotoxic alkaloids that dominate safety discussion of comfrey, coltsfoot and borage are a Boraginaceae, Asteraceae and Fabaceae concern. They are not a Saururaceae concern and are not reported from this plant.
• An established hepatotoxicity signal. Unlike several herbs with genuine case-report literature for liver injury, this plant does not have a recognised drug-induced-liver-injury profile. That is a statement about the absence of a signal, not a guarantee.
• A hydroquinone-style duration cap. Nothing in its chemistry creates the situation seen with uva ursi, where the active moiety itself is the reason for a strict time limit.
• Thiaminase activity. The vitamin B1-destroying enzyme that justifies caution with horsetail is not a described property here.
• Stimulant-laxative anthraquinones. The traditional use for constipation is not senna-style purgation; no anthraquinone laxative constituents are described.
• A coumarin-based bleeding risk. Its flavonoids are not coumarins, and in any case the popular inference is wrong twice over — plain coumarin is not an anticoagulant, dicoumarol from spoiled sweet clover is. No anticoagulant mechanism is described for this plant. That said, no interaction study exists either, so the absence of a mechanism is not the same as a tested reassurance.
• Carcinogenicity. No carcinogenicity finding attaches to correctly identified H. cordata. The cancer concern in this space belongs to aristolochic acid, which the plant does not appear to make.

Key Research Papers

Cited as PubMed searches rather than fixed identifiers, so a reader lands on the live record even where a secondary source has a detail wrong. Where metadata could not be verified, a topic search is linked rather than a citation asserted.

1. Analysis of aristolochic acids in Houttuynia cordata by liquid chromatography–tandem mass spectrometry — Journal of Mass Spectrometry, 2021. Aristolochic acids I and II absent or below a low nanogram-per-gram detection limit across twenty fresh and sun-dried samples. The single most important paper on this page. PubMed search.
2. Safety evaluation of aristolactams from the edible herb Houttuynia cordata — Food Research International, 2025. The mouse study comparing aristolactams against aristolochic acid I, with the pharmacokinetic explanation for the difference. PubMed search.
3. Rapidly progressive interstitial renal fibrosis in young women: association with slimming regimen including Chinese herbs — The Lancet, 1993. The Belgian case series, and the origin of the whole field. PubMed search.
4. Urothelial carcinoma associated with the use of a Chinese herb (Aristolochia fangchi) — New England Journal of Medicine, 2000. The cancer follow-up to the Belgian cohort. PubMed search.
5. Aristolochic acid and the etiology of endemic (Balkan) nephropathy — PNAS, 2007. Links the Balkan disease to dietary contamination with Aristolochia clematitis. PubMed search.
6. Aristolochic acid-associated urothelial cancer in Taiwan — PNAS, 2012. Mutational-signature evidence connecting exposure to upper urinary tract cancer at population scale. PubMed search.
7. Aristolochic acid DNA adducts and the A-to-T mutational signature — topic search on the nitroreduction mechanism that aristolactams, lacking the nitro group, do not enter. PubMed search.
8. Aristolactam nephrotoxicity — topic search on the compound class itself, including the cell-culture findings that run in the less reassuring direction. PubMed search.
9. Asarum and aristolochic acid content — topic search on the within-family case that shows the hazard tracks family and genus rather than order. PubMed search.
10. DNA barcoding for herbal product authentication — topic search covering ITS2, rbcL and psbA-trnH markers, and why a marker-compound assay cannot substitute for them. PubMed search.
11. Adulteration and substitution in herbal supply chains — topic search on the mechanism behind every documented aristolochic acid episode. PubMed search.
12. Acute and subacute toxicity evaluation of Houttuynia cordata ethanol extract in rats — Journal of Applied Toxicology, 2021. Oral rodent toxicology; note the material is an ethanolic extract, not a tea and not the injection. PubMed search.

External Resources

• IARC Monographs — the WHO cancer agency's classifications, including for aristolochic acid.
• LiverTox (NIH) — the reference on herb-induced liver injury, for checking whether a plant has an established hepatotoxicity record.
• NIDDK — Kidney Disease — the US National Institute of Diabetes and Digestive and Kidney Diseases patient information.
• World Flora Online — accepted names and family placement, for checking taxonomy claims like the ones on this page.

Connections

• All Herbs
• Houttuynia cordata (Fish Mint) — the main topic page.
• Houttuynia Benefits Deep Dive — the hub for this section.
• Injectable Reactions and Why Route Matters — the separate, and much better documented, safety subject.
• Houttuynia as Food and in Traditional Medicine — what long dietary use does and does not settle.
• Which Tinospora, and Adulteration — the site's other family-substitution case, with the same marker-versus-identity problem.
• Tinospora Benefits — the hub for that case.
• Kidney Disease — the organ at the centre of the aristolochic acid story.
• Acute Kidney Injury — how the Belgian cases presented.
• Heavy Metals — the other contamination risk that attaches to unverified imported herbal products.
• Perilla Ketone, Species and Safety — a neighbouring herb whose safety question also turns on which plant you actually have.
• Vietnamese Coriander — the other Vietnamese raw herb, for comparison of evidence depth.

Back to Table of Contents