The dangerous ones

What it is

The plants in traditional use that are known to cause serious harm — not through interaction with something else, but on their own.

This article exists because a section describing what plants have traditionally been used for, without naming the ones that damage organs, would be a herbal with the dangerous pages torn out.

Where it came from

The best-documented case is Belgian, and it is worth telling properly because it shows how the harm actually happens.

From 1990 a Brussels weight-loss clinic added two Chinese herbs to its regimen. One was meant to be Stephania tetrandra. What arrived from the supplier was Aristolochia fangchi.

The two plants have similar Chinese names — han fang ji and guang fang ji — and the substitution was not caught by anyone in the chain. Within a few years young women who had been through the clinic were arriving in renal units with rapidly progressive kidney failure. Around a hundred cases were identified. Many required dialysis or transplant, and follow-up found urinary tract cancers in a substantial share of those tested.

The pattern was then recognised elsewhere. Balkan endemic nephropathy, a kidney disease known in villages along the Danube since the 1950s, was eventually traced to Aristolochia clematitis seeds contaminating local wheat. And in Taiwan, where aristolochic acid preparations were widely used before a 2003 ban, population studies found elevated rates of the same cancers.

Aristolochic acid is now classified as a Group 1 human carcinogen by the IARC — the same category as tobacco and asbestos. There is no known safe dose.

How it is used

Traditionally, for a range of complaints, in several systems. That is precisely the point: a long history of use is not evidence of safety, because the harm here takes years to appear and does not look like poisoning when it does.

What we can and cannot say

We can name the ones with the strongest evidence of harm.

• Aristolochia — kidney failure and urothelial carcinoma, as above. Banned or restricted in most jurisdictions including Australia. Still appears in imported preparations.
• Pyrrolizidine alkaloids, found in comfrey, borage, coltsfoot and butterbur — cumulative liver damage including veno-occlusive disease. Comfrey has caused deaths, and internal use is restricted in many countries while it remains available as a topical product.
• Ephedra (ma huang) — a genuine sympathomimetic. Associated with strokes, arrhythmias and deaths, and banned as a supplement ingredient in the US in 2004 after around 150 reported deaths, including a well-publicised one in professional baseball.
• Aconite (monkshood, fu zi) — one of the most acutely toxic plants in medicinal use anywhere. Cardiotoxic, with a narrow margin between the traditional processed dose and a fatal one. Deaths are reported regularly, mostly from inadequate processing.
• Kava — hepatotoxicity, including cases requiring transplant. The picture is genuinely contested: the traditional Pacific aqueous preparation from the root has a far better safety record than the Western acetone and ethanol extracts, and some of the implicated products used aerial parts of the plant, which are not what Pacific practice uses. Several national bans were later relaxed.
• Green tea extract — now a leading identified cause of herb-induced liver injury in Western case series. Drinking green tea is not the issue; concentrated extract capsules are.
• Black cohosh — liver injury reports sufficient for regulators including the TGA to require a warning statement.
• Yohimbe — hypertension, arrhythmia and anxiety, with wide variation in actual alkaloid content between products.

We can say the mechanisms of harm sort into four kinds, and this is more useful than any list:

Intrinsic toxicity — the plant is dangerous at doses people actually use. Aconite, aristolochia.

Cumulative toxicity — each dose is fine and the total is not. The pyrrolizidine alkaloids work this way, which is why "I've taken it for years with no problem" is not the reassurance it sounds like.

Misidentification — the Belgian cases. The plant that arrives is not the plant that was ordered, and nobody in the chain is checking.

Dose variability — the same species, prepared differently, can differ in active content by an order of magnitude. Growing conditions, harvest time, plant part, and extraction solvent all move it. A pharmaceutical tablet states its content and is tested against that statement. A herbal preparation generally does neither.

We can say the reporting is systematically incomplete. Adverse events from herbal products are under-reported because the person having them often does not connect the two, does not mention the supplement to the clinician, and is not asked. Liver injury in particular is usually painless until it is advanced. So the case series in the literature are a floor, not an estimate.

And we can say what this article is not arguing. It is not that plants are dangerous and drugs are safe — pharmaceuticals have their own long list, kept and published precisely because the industry is required to keep it. The argument is narrower and harder to dodge: the plants here are dangerous for the same reason the useful ones are useful, and a tradition that claims the second half has to own the first.

We cannot tell you what is safe. What can be said is that a long history of traditional use tells you a plant was widely used, not that it was widely survived, and that the two are different claims.

Further reading

• Vanherweghem et al. on the Belgian nephropathy cases (The Lancet, 1993).
• The IARC monograph on aristolochic acids.
• LiverTox, the NIH database of drug and herbal liver injury — free, searchable and unsensational.