There’s one thistle that has a place in hospital medicine, and it’s probably not doing what you’d guess.

Milk thistle, Silybum marianum, has white veins running across its leaves. The legend says they’re drops of the Virgin Mary’s milk, spilled while she nursed under its leaves on the way out of Egypt, and that’s where both the common name and the species name come from. Its seeds are also the source of one of the best-known herbal supplements there is, sold for “liver support” in capsules on pharmacy shelves. I’m leaving that aside. What I’m interested in is a narrower, stranger use: an intravenous drug made from a compound in those seeds, silibinin, which doctors in Europe give to people who have eaten a death cap mushroom.

This part surprised me. As far as I can tell, the drug doesn’t neutralise the poison. It closes a door.

The death cap’s toxin, amatoxin, is cruel mostly because of its timing. For the first six hours or more after the meal, nothing happens. Then there’s a day or so of violent vomiting and diarrhoea, and then, often, the person seems to get better. The name for that stage is “false recovery.” Meanwhile the toxin keeps doing its damage out of sight. The liver pulls amatoxin out of the blood through a transporter in its cell membranes, a protein called OATP1B3, and once inside, the toxin kills the cell. Some of it also gets excreted into bile, which collects in the gallbladder. Then, when the person eats, the gallbladder squeezes, the bile goes back into the gut, the toxin is reabsorbed, and the liver takes it up again. The poison goes round in a loop, and the organ that ought to be clearing it keeps letting it back in.

Silibinin blocks that transporter. It doesn’t destroy anything. It makes the liver cell refuse re-entry, so toxin that would have gone round again has a chance to leave through the kidneys and gut instead. The treatment isn’t a counter-poison. It works by stopping the liver from taking the toxin back in.

The thing I most want to be honest about is how uncertain the evidence was, and still is. Nobody has ever run a randomised controlled trial of silibinin for death cap poisoning, and it’s hard to see how anyone could. Cases are rare, scattered, and urgent, and you can’t ethically give half of a group of poisoned people a placebo when the thing they need has been used for decades. So for most of that time the case rested on case series. The most-quoted figure, nearly 1,500 documented cases in which patients given silibinin died less often than those given the older treatments, comes from a 2012 review with an author from the company that makes the drug. A 2022 review with no such tie, pooling 877 published cases, points the same way: about 90% of patients given silibinin survived, against 59% of those who got supportive care alone. That’s real information, but it isn’t the same as proof. Who got which drug wasn’t random, and published cases lean toward the ones worth writing up.

In the United States the drug still isn’t approved. For years the way to get it has been to enrol the patient, one case at a time, in a single open trial run under an FDA investigational-drug permit. That trial exists because of Todd Mitchell, a doctor in Santa Cruz. In January 2007, six members of one family came into his hospital after eating death caps. He found the European reports, got emergency permission for the drug, and went on to lead an open, single-arm trial. It was registered in 2009 and enrolled patients slowly, a few a year, for well over a decade.

Results from that trial appeared this March, as a preprint, which means they haven’t been through peer review yet. Of the 99 patients who got the treatment, 87 recovered without a liver transplant, six needed a transplant, and six died. The authors also report a much better number, 85 of 86, for the patients whose care followed the protocol without interruption. I’d be careful with that one. Filtering by “who stayed on protocol” can quietly select for people who were going to do better anyway, and a single-arm study has no comparison group to check it against.

The part I found most interesting isn’t the headline figure, though. The trial wasn’t testing silibinin alone. It paired the drug with steady, uninterrupted IV fluids to keep the kidneys clearing the toxin, fasting so the gallbladder wouldn’t squeeze its load back into the gut, and a drug called octreotide to suppress the same squeeze. When the authors looked at what best predicted recovery, the strongest factor wasn’t the thistle. It was uninterrupted hydration. Earlier silibinin came second. The authors put their framing in the preprint’s original title, elimination drives recovery, and a revised version sharpens it to elimination capacity determines outcome. The whole protocol is about keeping the way out open and the way back in shut.

So the thistle’s part in this, if the data holds up, is real but secondary. It’s one door among several. I like it better that way than as a miracle. A plant named after the milk of a nursing mother turns out to be useful in a poisoning for what it keeps out, not for anything it gives, and only if the fluids keep running.

One plain note, because it matters: none of this is about the capsules. The hospital drug is a modified, water-soluble form of silibinin, given into a vein in an ICU. Nobody should read this and treat a suspected mushroom poisoning at home. If someone may have eaten a wild mushroom they can’t identify with certainty, that’s a call to poison control or emergency services right away, even if they feel fine. The first few hours of feeling fine are part of how this toxin works.

— Briar