Caffeine

Why this poison is interesting

Caffeine is the most widely consumed psychoactive drug on earth, and its overdose is one of the more dangerous in this band — a combination that makes it easy to underestimate. The reason the danger is real is on the label of its licensed form: caffeine is structurally related to the methylxanthines theophylline and theobromine¹, and in overdose it stops behaving like a cup of coffee and starts behaving like theophylline. Everything that makes theophylline one of the nastier cardiovascular poisons — the catecholamine storm, the low potassium, the tachyarrhythmias, the seizures — caffeine reproduces.

At ordinary doses caffeine's effects are attributed to antagonism of adenosine receptors, both A₁ and A₂A subtypes¹ — it removes the brake adenosine puts on arousal and cardiac activity. At toxic concentrations two further mechanisms come into play: inhibition of phosphodiesterase (raising intracellular cyclic AMP) and a massive release of catecholamines. The result is a β-adrenergic storm: tachycardia and tachyarrhythmia, hypotension from β₂ vasodilatation, and the metabolic signature of β₂ stimulation — hyperglycaemia, hypokalaemia¹ and a lactic acidosis. This is not a stimulant's jitters; it is the physiology of an adrenaline infusion the patient cannot switch off.

The modern hazard is availability. Caffeine is sold not only as tablets but as bulk powder and concentrated pre-workout preparations, and a teaspoon of pure powder contains a dose that has killed. Poison-centre data record clusters of fatalities against a background of tens of thousands of benign exposures², and the deaths share a feature that defines the management problem: the abrupt onset of intractable seizures needs to be anticipated.² The gap between how safe caffeine feels and how a gram-scale overdose behaves is the whole reason it earns a monograph.

A poison is a drug whose kinetics have escaped its pharmacology.

The toxic principle

The toxic principle is the parent methylxanthine, acting through three mechanisms whose balance shifts with concentration. At therapeutic levels the dominant action is antagonism of adenosine receptors¹: adenosine normally promotes sleep, slows the heart and restrains catecholamine release, and blocking it produces alertness and mild cardiac stimulation. This is the coffee-cup pharmacology, and on its own it is benign.

At toxic concentrations two mechanisms that are trivial at therapeutic doses become important. Caffeine inhibits phosphodiesterase, raising intracellular cyclic AMP and amplifying β-adrenergic signalling; and it provokes a large release of endogenous catecholamines. The combination reproduces the effect of an adrenaline excess, and the clinical and biochemical picture follows directly: tachyarrhythmia, tremor, and the β₂-adrenergic metabolic triad of hypokalaemia (potassium driven into cells), hyperglycaemia (glycogenolysis) and a lactic acidosis. The licensed-form overdose data list hyperglycaemia, hypokalaemia, fine tremor… seizures, tachycardia, vomiting¹ — the fingerprint of exactly this mechanism. Inferred

Toxicokinetics

Caffeine's kinetics are those of a small, freely distributed, hepatically cleared stimulant, and two features shape the overdose: rapid and complete absorption, so the storm arrives quickly, and enterohepatic recirculation, which gives repeated-dose charcoal a rationale. It is a good extracorporeal target when the concentration is high enough to justify it.

Metabolism and the metabolites

Caffeine is demethylated in the liver, chiefly by CYP1A2, to paraxanthine and — tellingly — to theobromine and theophylline. None of the metabolites is the toxic species: the poisoning is the parent methylxanthine's, and the metabolic route matters mainly because it is saturable and because it is inhibited by common drugs and by pregnancy, both of which prolong the half-life.

Elimination and accumulation

Elimination is metabolic and, in overdose, slower than the textbook half-life suggests, because the clearance route saturates. A therapeutic caffeine half-life of a few hours lengthens as the concentration rises and lengthens further where CYP1A2 is inhibited — by fluvoxamine, by ciprofloxacin, by oral contraceptives, and markedly by pregnancy and liver disease. The accumulation that matters is not renal but hepatic-metabolic: the same dose is far more dangerous in a slow metaboliser.

There is no organ that caffeine slowly destroys; the accumulation question is the persistence of a stimulant with a saturable clearance and a recirculating pool. In severe poisoning, where supportive care and charcoal are not keeping pace, extracorporeal removal shortens the exposure directly.¹ Inferred

Target organs — and why those

Timeline of effects

What the mechanism predicts at the bedside

• Treat it as theophylline. The mechanism, the electrolytes and the management overlap almost completely; the theophylline page is the closest reference for a severe caffeine ingestion.
• Anticipate the seizure. Poison-centre experience is explicit that the abrupt onset of intractable seizures needs to be anticipated²; benzodiazepines are first-line and should be ready before they occur.
• Expect a low potassium, and replace it cautiously. The hypokalaemia is redistributive¹; over-replacement risks rebound hyperkalaemia as the caffeine clears. Correcting the β₂ drive corrects the potassium.
• The hypotension may respond to a β-blocker. Because it is β₂-mediated vasodilatation, a short-acting agent such as esmolol can raise the pressure and control the tachyarrhythmia — the same paradox as theophylline. See beta-blockers for the receptor logic in reverse.
• Give activated charcoal, and consider repeated doses. Caffeine is well adsorbed and undergoes enterohepatic recirculation, so multiple-dose charcoal is an elimination strategy, not just decontamination.¹ See activated charcoal.
• Consider haemodialysis for severe poisoning. A very high concentration with refractory arrhythmia or seizures is dialysable; exchange transfusion also lowers the level.¹
• Ask about the preparation. Bulk powder and pre-workout concentrates put a lethal quantity within a teaspoon; a tablet count understates the risk of a powder ingestion.
• Watch the lactate and glucose. A lactic acidosis and hyperglycaemia are part of the β-adrenergic picture¹, not evidence of a second diagnosis.

The antidote, from the poison's side

There is no specific antidote. Management is aimed squarely at the β-adrenergic storm and its consequences, and the interventions read across almost unchanged from the theophylline page: control the sympathetic drive, protect against the arrhythmia and the seizure, and remove the drug where the concentration justifies it.

β-blockade (e.g. esmolol)

The mechanistically targeted treatment for the cardiovascular storm: a short-acting β-blocker slows the tachyarrhythmia and, by removing β₂ vasodilatation, can raise the blood pressure — the same counterintuitive move used for theophylline. Short-acting so it can be stopped if tolerated poorly. Not a drug monograph in this estate; see beta-blockers for the receptor pharmacology.

Benzodiazepines

First-line for the seizures, which are anticipated in any significant ingestion², and for agitation — a GABA-ergic brake against a cortex the adenosine blockade has disinhibited.

Activated charcoal (including multiple doses)

Well adsorbed, and — because caffeine recirculates enterohepatically — repeated doses actively remove absorbed drug, not just unabsorbed drug.¹ See activated charcoal.

Potassium replacement — cautious

For the redistributive hypokalaemia¹, given carefully because the potassium rebounds as the caffeine clears. Correcting the β₂ drive is the more physiological correction.

Haemodialysis / exchange transfusion

For severe poisoning — very high concentration, refractory arrhythmia or seizures. Caffeine is a good dialysis target, and exchange transfusion lowers the level in neonates.¹ There is no formal EXTRIP recommendation, so the decision is clinical.

Critical appraisal

• The symptom list in the glance is neonatal in origin. The specific overdose features — hyperglycaemia, hypokalaemia, fine tremor… seizures¹ — come from the caffeine citrate label, which describes poisoning in preterm infants. The mechanism is the same in adults and the adult fatalities show the same physiology²⁴, but the itemised list is a neonatal one and is cited as such.
• The β-adrenergic mechanism is Inferred for the cardiovascular and CNS effects and Established for the electrolytes. That β₂ stimulation redistributes potassium and mobilises glucose is established physiology; the attribution of the arrhythmia, the hypotension and the seizure threshold specifically to catecholamine release and adenosine blockade is the standard and well-supported inference, badged as inference rather than as demonstrated in poisoned humans.
• The β-blocker-for-hypotension claim is mechanistic and extrapolated from theophylline. The reasoning — β₂-mediated vasodilatation reversed by β-blockade — is sound and is applied to theophylline in practice; the direct caffeine evidence is limited, and the page badges it Inferred and cross-links the receptor logic rather than asserting a trial.
• Adult pharmacokinetic figures are textbook values. The ~0.5 L/kg volume, ~36% protein binding, CYP1A2 route and ~4–6 h half-life are standard pharmacology quoted as such; the overdose-relevant point — that clearance saturates and is inhibited in pregnancy and by CYP1A2 inhibitors — is the one the page relies on.
• No lethal dose is given. That a purchasable quantity of powder can be fatal is stated as a hazard, not as a dose; the fatality literature²³ documents deaths without this page reproducing a lethal amount, and risk assessment belongs to TOXBASE and NPIS.
• The dialysis recommendation is clinical, not EXTRIP-graded. No EXTRIP recommendation exists for caffeine; the page states that it is dialysable and reserved for severe poisoning, which is a physicochemical and pragmatic judgement rather than a graded one.

References

1. 1
   Peyona 20 mg/ml solution for infusion and oral solution (caffeine citrate) — Summary of Product Characteristics. electronic medicines compendium, product 4098 (Chiesi). Sections 4.9 (Overdose) and 5.1 (Pharmacodynamic properties, mechanism of action). Overdose plasma caffeine 50–350 mg/L; reported features in preterm infants include hyperglycaemia, hypokalaemia, tremor, seizures, tachycardia; concentrations fall after exchange transfusion. medicines.org.uk/emc/product/4098
2. 2
   Shum S, Seale C, Hathaway D, et al. Acute caffeine ingestion fatalities: management issues. Veterinary and Human Toxicology 1997 Aug;39(4):228–230. PMID 9251173. Two deaths from caffeine overdose against a background of thousands of benign exposures; the abrupt onset of intractable seizures must be anticipated.
3. 3
   Yamamoto T, Yoshizawa K, Kubo S, et al. Autopsy report for a caffeine intoxication case and review of the current literature. Journal of Toxicologic Pathology 2015 Jan;28(1):33–36. PMID 26023259. A fatal caffeine intoxication with organ caffeine concentrations and a review of the small published fatal literature.
4. 4
   Han K, You KM, Lee H, et al. A case of refractory ventricular fibrillation after caffeine poisoning successfully treated by supportive care. Toxicology Reports 2022;9:1710–1712. PMID 36561958. Refractory ventricular fibrillation following caffeine poisoning, cited for the severity of the cardiac endpoint and recovery with supportive care.