Transcript
Paracetamol, also called acetaminophen, is one of the most widely used drugs in the world. It is a standard first-line treatment for fever and mild to moderate pain, and it has been in clinical use for more than a century. Yet a 2021 review in the journal Temperature, by the pharmacologist Samir Ayoub, opens by saying that its precise mechanism of action remains a mystery.
This episode explains why a drug this familiar still resists a full explanation. We will go through the main theories and what research has shown for each, including a 2025 study that put a new candidate on the table. Then we will turn to the other half of paracetamol's story: why too much of it can destroy the liver, why the damage arrives days later, and why the antidote has to be given early. This is general education about published research, not personal medical guidance.
First, why the question is hard. Most drugs have a simple story: a molecule binds a specific target, the target does something, and a symptom improves. Paracetamol does not fit that pattern.
It behaves differently from the drugs it is usually grouped with. Aspirin, ibuprofen and diclofenac are non-steroidal anti-inflammatory drugs, or NSAIDs, which work by blocking enzymes called cyclooxygenases, or COX enzymes. These enzymes make prostaglandins, chemical messengers that drive pain, fever and inflammation. Paracetamol relieves pain and lowers fever like those drugs, but its anti-inflammatory activity is minimal. It will not reduce the swelling in an arthritic joint, and it does not cause the stomach ulcers that NSAIDs can. A drug that acts like an NSAID for pain and fever but not for inflammation is already a puzzle.
There is a second difficulty. The places where paracetamol is thought to act, the brain, the spinal cord and the nerve endings that sense pain, are very hard to study directly in a living person. Most of the mechanistic evidence comes from test tubes, isolated cells, and rodents, and translating that to humans is where the arguments start. So there are several suspects, and none has been caught fully in the act.
The oldest and most prevailing suspect is COX inhibition, mainly in the central nervous system, the brain and spinal cord, rather than in inflamed tissue elsewhere in the body.
The first gap is that in classic laboratory screens, using ground-up cells and purified enzymes, paracetamol looks like a feeble COX blocker, even at concentrations far higher than anything that reaches the blood. In a test tube, it barely works. In a person, it clearly does.
One proposed way out is that paracetamol acts as a reducing agent. COX enzymes need a small amount of peroxide, a reactive oxygen compound, to switch themselves on, and paracetamol may interfere with that activation. The proposal is that this works only where peroxide levels are low, as in the brain. In inflamed tissue, where peroxide levels are high, paracetamol cannot compete, which would explain why it lacks the anti-inflammatory punch of ibuprofen.
There is also some human evidence on this point. In a 2008 study led by Burkhard Hinz, volunteers took a one-gram dose, and their blood was then tested outside the body. COX-2 activity in that blood was cut by more than 80 percent, comparable to NSAIDs. Not everyone accepts that this adds up to a clinically meaningful effect, but it shows that at an ordinary dose, in people, paracetamol can block COX far more than the test-tube screens suggest. So the same drug looks weak or strong depending on how you test it. That is the first and oldest theory: plausible, partly supported in people, and still not a complete explanation, because blocking COX does not by itself explain why paracetamol spares inflammation so thoroughly.
The second suspect had its moment and then collapsed. In 2002, researchers found a variant of the COX-1 enzyme in the brains of dogs and called it COX-3. It looked like the long-sought answer: a brain-specific COX enzyme that paracetamol might preferentially block.
It did not survive scrutiny. Several research groups found that in human and rodent tissue, the variant comes out as a truncated protein with no enzyme activity, so it could not be the explanation in people. The idea was abandoned in the research literature, but it still turns up in some older textbooks and websites. If you read an explanation that presents COX-3 as settled, treat it with caution.
The third suspect is the most intriguing, and it comes from a different angle. Instead of asking what paracetamol does, it asks what the body turns paracetamol into.
Paracetamol behaves in part as a prodrug, meaning a compound that is converted into an active substance after it is taken. In the liver, it loses an acetyl group, leaving a compound called 4-aminophenol. In the brain, an enzyme called FAAH joins 4-aminophenol to arachidonic acid, a fatty acid the body already uses. The result is a molecule called AM404. It was identified as a paracetamol metabolite by a Swedish team led by Edward Högestätt in 2005.
AM404 touches the pain system in several ways. It acts on TRPV1, a receptor involved in sensing heat and pain, and weakly on CB1, one of the body's cannabinoid receptors, the same system that cannabis acts on. It also blocks the uptake of anandamide, one of the body's own cannabinoid-like messengers. In rodents, switching off FAAH, or TRPV1 in the brain, or blocking CB1 receptors, reduced paracetamol's pain relief in several experiments. And in 2017, researchers reported finding AM404 in the spinal fluid of human patients who had been given paracetamol, the first evidence that the conversion happens in people.
Now the qualifiers. The amounts found are very small, and a key open question is whether AM404 reaches pain-relieving concentrations in humans. Some animal evidence also points elsewhere. In mice lacking CB1 receptors, and in mice lacking TRPV1 receptors, paracetamol still lowered body temperature just as it did in normal mice. If this pathway were responsible for that effect, removing its targets should have blunted it. That is one effect in one species, so it does not rule AM404 out for pain relief, but it means its role is contested. Ayoub's own earlier work points instead toward a variant of COX-1 for the fever-related effects.
Then, in 2025, AM404 came back with a twist. A team at the Hebrew University of Jerusalem, led by Alexander Binshtok and Avi Priel, reported in the Proceedings of the National Academy of Sciences that AM404 is also made in the pain-sensing nerve endings themselves, outside the brain. There, it directly blocks two sodium channels, called NaV1.7 and NaV1.8, which pain-sensing nerves need in order to fire. AM404 binds at the same site that local anaesthetics use. Other paracetamol metabolites did not do this, and in rats AM404 reduced pain-related behaviour.
Some headlines declared the mystery solved. That goes further than the study. It was done in rat neurons and rats, not people, and it adds a peripheral route alongside the central ones rather than replacing them. But it shows how alive the question still is: the same metabolite now has a proposed role in the brain and a separate one at the nerve endings.
The fourth suspect is serotonin. In animal models, blocking certain serotonin receptors, specifically 5-HT1A, 5-HT3 or 5-HT7, reduces or abolishes paracetamol's pain relief. That sounds like a smoking gun. But paracetamol does not bind to serotonin receptors itself. One interpretation is that the serotonin effect is a downstream consequence: change prostaglandin or cannabinoid signalling in the brain, and the descending serotonin pathways that dampen pain signals change too. So the pathway matters for the effect, but it may be a consequence rather than the cause.
To recap the mechanism section: central COX inhibition explains the drug's resemblance to NSAIDs, and has some support in human blood, but looks weak in classic screens. COX-3 was an elegant answer that did not hold up in humans. AM404 offers routes through the pain and cannabinoid systems in the brain, and now possibly through sodium channels at the nerve endings, but how much it contributes in people is unknown. Serotonin pathways are clearly involved, but probably downstream. Each theory explains something. None explains everything.
One more point bears on how much there is to explain. Paracetamol's effect is real but modest. In most trials of dental pain, such as after wisdom-tooth removal, ibuprofen does better. And in large reviews, paracetamol was no better than placebo for low back pain, with only a small benefit for hip and knee osteoarthritis. So the drug is both universally used and unevenly effective, and a complete mechanism would have to account for that too.
Now the second half: the liver. Taken as directed, paracetamol is handled safely. When a normal dose reaches the liver, most of it is neutralised by attaching other molecules to it, a process called conjugation, mainly with glucuronide or sulfate. These conjugates are harmless and leave the body in urine. Roughly 85 to 90 percent of a dose follows this route.
A small share, around 5 to 10 percent, takes a different path. Liver enzymes, mainly one called CYP2E1, turn it into a reactive, toxic compound called NAPQI, short for N-acetyl-p-benzoquinone imine.
Even that small share is normally not a problem, because the liver holds a protective supply of a molecule called glutathione. Glutathione grabs NAPQI and neutralises it, and the product is excreted harmlessly. The system works as long as the glutathione supply keeps up with the NAPQI being produced.
An overdose breaks that balance. With a very large dose, the safe conjugation routes become saturated, so more of the drug is pushed through the NAPQI route. More NAPQI is made, glutathione is used up, and once it is depleted, NAPQI is free to bind to proteins inside liver cells, especially in the mitochondria, the cells' power plants. That binding damages the cells, and the result can be acute liver failure.
The scale of the problem is real. Paracetamol is the leading cause of acute liver failure in many countries, including the United States, the United Kingdom and Australia. The usual maximum for adults is 4,000 milligrams in 24 hours. That limit has a safety margin built in, but the margin is narrower for people who drink heavily, are malnourished, or take paracetamol at high doses for several days.
Why is the damage delayed? Because the binding of NAPQI is only the first step. Damaged mitochondria start producing reactive oxygen, stress signals inside the cell amplify the injury, and eventually the mitochondria fail and the liver cells die. That cascade unfolds over hours to days. In the first day after an overdose, a person may feel only mildly unwell, with some nausea, or nothing at all. Liver blood tests then climb, and in serious cases the damage peaks around three to four days after the overdose. This is what makes paracetamol poisoning so treacherous: by the time someone feels seriously ill, much of the damage is already done.
The antidote is N-acetylcysteine, usually called NAC, given as an intravenous infusion over many hours. It supplies the building block the liver needs to make new glutathione, so that NAPQI can be neutralised before it binds to proteins. That is why timing matters so much. Given early, NAC stops the cascade before it starts. Once NAPQI has already bound and the mitochondrial damage is under way, restoring glutathione cannot undo it. Australian and New Zealand guidelines published in 2020 note that when NAC is started within eight hours of an overdose, the risk of liver injury is low, under one percent. NAC is still given after eight hours, and it still helps, but the protection is less complete.
To restate the overdose chain: a normal dose is mostly handled safely, with only a small share becoming NAPQI, and glutathione neutralises that share. In an overdose, the safe pathways saturate, more NAPQI forms, glutathione runs out, and the toxic compound sets off a slow cascade that kills liver cells. NAC restores the protective supply, and works best before that cascade begins.
The practical takeaway follows directly from that. Many cold and flu products, and many combination painkillers, also contain paracetamol, so it is easy to double up without realising. And because an overdose can feel like very little at first, feeling fine is not a reason to wait. If you think someone has taken too much, get advice straight away, even if they seem well. In Australia, the Poisons Information Centre is on 13 11 26, any time of day or night. In an emergency, call 000.
So what is the answer to the guiding question? Paracetamol's mechanism is elusive because it does not behave like a single-target drug. It likely acts through several linked routes: COX inhibition under low-peroxide conditions in the brain, a metabolite that touches pain and cannabinoid receptors centrally and may block sodium channels at the nerve endings, and serotonin pathways that probably follow downstream. Each has supporting data and each has a gap, and the hardest part, measuring what happens at those sites in living people, is exactly where the evidence is thinnest.
Several questions remain open. Which COX variant, if any, does paracetamol inhibit in the brain, and how? Does AM404 reach meaningful concentrations in people, and does the newer sodium-channel finding hold up in humans? How do serotonin effects arise without direct binding? After more than a century of use, the most familiar painkiller in the world is still, in important ways, an open question. The danger of too much is the part we understand best.