Theophylline

Why this poison is interesting

Most pharmaceutical overdoses in this library make the patient sleepy. Theophylline does the opposite, and it does so by borrowing the patient's own physiology. It is a methylxanthine — chemically a caffeine relative — and in overdose it produces a picture indistinguishable from a catecholamine crisis: tachycardia, tremor, vomiting, hypokalaemia, hyperglycaemia, lactic acidosis, agitation and, at the extreme, arrhythmia and seizures. The striking thing is that theophylline supplies none of that adrenergic drive directly. It releases the patient's own adrenaline and noradrenaline, and simultaneously removes the endogenous brake — adenosine — that would otherwise damp the same systems down.¹³

• Adenosine antagonism. Theophylline is a non-selective adenosine-receptor blocker. Adenosine is an endogenous anticonvulsant and a negative chronotrope; blocking it lowers the seizure threshold and speeds the heart. Established
• Phosphodiesterase inhibition and catecholamine release. Circulating adrenaline and noradrenaline rise markedly in theophylline toxicity, and β-adrenergic stimulation drives potassium and glucose into cells, stimulates glycogenolysis and lipolysis, and produces the tachyarrhythmias. Inferred
• A narrow therapeutic index. The label puts the ceiling of the therapeutic range at 20 mg/L and says plainly do not exceed it;² toxicity begins just above the top of the target range, so there is almost no distance between a therapeutic concentration and a toxic one.

A poison is a drug whose kinetics have escaped its pharmacology — and theophylline's pharmacology is a hormone the patient makes for themselves.

The toxic principle

The poison is theophylline, unchanged. Aminophylline — theophylline complexed with ethylenediamine to improve solubility for injection — delivers the same molecule, so an aminophylline overdose is a theophylline overdose with a different label on the ampoule. There is no activation step and no toxic metabolite; the drug that treats asthma at 10 mg/L is the drug that causes seizures at 100. Established

It acts through at least three mechanisms, and in overdose they stop being separable — they converge on one adrenergic, hypermetabolic state.

• Non-selective adenosine-receptor antagonism. This is the mechanism most specific to overdose. Adenosine slows the sinoatrial and atrioventricular nodes and raises the seizure threshold; blocking it removes both protections at once — which is why the arrhythmias and the seizures track together. Established
• A rise in circulating catecholamines, with the whole β-adrenergic syndrome that follows: tachyarrhythmia, the transcellular potassium shift, hyperglycaemia, and skeletal-muscle tremor. Inferred
• Phosphodiesterase inhibition, raising intracellular cyclic AMP. This is probably a minor contributor at therapeutic concentrations but adds to the adrenergic picture in overdose. Inferred

Toxicokinetics

Metabolism and the metabolites

As with salicylate, this is the section where the interesting thing is an absence. Theophylline's hepatic metabolites — 1,3-dimethyluric acid and 3-methylxanthine — are not the poison.² The parent drug does all the damage, so there is no metabolic branch to block and no cofactor to replace.

What changes the answer

• Anything that inhibits CYP1A2 turns a therapeutic dose toxic without an overdose. Ciprofloxacin, erythromycin, fluvoxamine, cimetidine and acute viral illness all cut clearance; the classic chronic-toxicity story is a stable patient given a new antibiotic. Established
• Smoking cessation raises the concentration, because tobacco smoke induces CYP1A2 and stopping removes the induction — a patient admitted and unable to smoke can drift up over days. Inferred
• Heart failure and liver disease reduce clearance, which is one reason the same population that is prescribed theophylline is the population that accumulates it. Inferred

Elimination and accumulation

Theophylline cannot be alkalinised out — the renal route is too small. What it can be is adsorbed back out of the circulation, and this is one of the few poisons where multi-dose activated charcoal has a genuine, mechanistic rationale rather than a hopeful one.

Why multi-dose charcoal actually works here

Theophylline diffuses back from the circulation across the gut wall down a concentration gradient, and repeated oral charcoal maintains a sink on the luminal side that keeps that gradient running — so charcoal removes drug that has already been absorbed, not merely drug still waiting in the gut. The SmPC states it directly: "Repeated doses of activated charcoal given by mouth can enhance theophylline elimination."² EXTRIP goes further and recommends that MDAC be continued even during dialysis (1D),¹ which only makes sense if the two are removing drug by independent routes. Established

The EXTRIP position

141 articles, 143 patients, 10 deaths; low-to-very-low quality of evidence for every recommendation, but the dialysability itself is graded level A — unusually strong for this workgroup.¹

• Recommended in severe theophylline poisoning (1C), and specifically for [theophylline] >100 mg/L (555 µmol/L) in acute exposure (1C), seizures (1D), life-threatening dysrhythmias (1D), shock (1D), a rising concentration despite optimal therapy (1D), or clinical deterioration despite optimal care (1D).¹
• Suggested in chronic exposure at >60 mg/L (333 µmol/L) (2D), or >50 mg/L (278 µmol/L) if the patient is under 6 months or over 60 years (2D) — the concentration thresholds are lower for chronic poisoning, which is the whole point.¹
• Suggested if gastrointestinal decontamination cannot be given (2D).¹
• Stop when clinical improvement is apparent or [theophylline] <15 mg/L (83 µmol/L) (1D).¹
• Intermittent haemodialysis preferred (1C); haemoperfusion (1C) or CRRT (3D) are alternatives; exchange transfusion in neonates (2D). Continue MDAC throughout (1D).¹

Target organs — and why those

Timeline of effects

Theophylline has two quite different clocks depending on whether the exposure is acute or chronic — and a third variable, the formulation, that can hide the first.

What the mechanism predicts at the bedside

Why one concentration cannot be trusted — for two separate reasons

In acute overdose, absorption may still be climbing behind a modified-release shell for up to twelve hours,² so an early number is a floor, not a ceiling. In chronic over-medication, the plasma concentration understates the body burden and the patient's frailty, so even a stable number can accompany major toxicity.³ The two failure modes are opposite in mechanism but identical in consequence: serial concentrations, read against the patient, are the only safe approach.

Why the potassium needs watching in both directions

The hypokalaemia is a transcellular shift under β-adrenergic drive, not a true deficit.² It must be corrected to protect the heart, but the potassium is intracellular and will return: the SmPC's own warning is that serious hyperkalaemia can develop in recovery if large amounts have been given.² Replace with the recovery in mind. Established

Why the seizures are a dialysis trigger, not just an anticonvulsant problem

Because the drug has removed the endogenous anticonvulsant (adenosine) and blunted the exogenous one (benzodiazepines).² Escalating anticonvulsants against a rising theophylline concentration is treating the symptom while the cause climbs. EXTRIP lists seizures as a stand-alone dialysis indication for exactly this reason.¹ Inferred

Why β-blockade is both rational and hazardous

Much of the toxicity is β-adrenergic, so a β-blocker is mechanistically the right antidote for the tachyarrhythmia and even the hypokalaemia — and the SmPC endorses it in extreme cases. But it withholds it from asthmatics,² who are precisely the people prescribed theophylline. It is a clean example of a mechanistically correct treatment defeated by the patient population it would be used in.

Why lidocaine is the wrong antiarrhythmic

The label warns against proconvulsant antiarrhythmic agents such as lignocaine (lidocaine) because of the risk of causing or worsening seizures.² In a poison that has already lowered the seizure threshold by antagonising adenosine, adding a proconvulsant drug to treat the arrhythmia trades one lethal endpoint for another. Established

The antidote, from the poison's side

There is no antidote in the antagonist sense — no molecule that binds theophylline, and adenosine itself is far too short-lived to be given as a countermeasure. What exists is removal, and theophylline is unusually good at being removed by two independent routes.

• Multi-dose activated charcoal is disease-modifying, not merely decontaminating. Because theophylline back-diffuses across the gut wall, repeated charcoal removes absorbed drug, and the SmPC and EXTRIP both endorse it — EXTRIP even continues it during dialysis.¹² It is one of the small number of poisons for which MDAC has a real mechanistic case. Established
• Dialysis is the definitive removal, and theophylline is close to the ideal substrate: small, water-soluble, minimally bound and confined to a small volume, which is why the dialysability is graded level A.¹ Unlike salicylate, the case does not rest on protein binding coming undone — theophylline was always available to the machine.
• The β-adrenergic manifestations can be opposed pharmacologically with a β-blocker where the patient is not asthmatic,² which is the closest theophylline has to a mechanism-directed antidote — countering the hormone excess the poison provokes rather than the poison itself.
• The dangerous interventions are the intuitive ones: potassium given without anticipating rebound, lidocaine for the arrhythmia, and escalating benzodiazepines for seizures that the drug has made benzodiazepine-resistant.² The mechanism predicts each of these traps.

Critical appraisal

1. The catecholamine-release mechanism is inferred from the clinical picture and from measured catecholamine rises, not proven at the receptor in human overdose. Inferred It explains the entire syndrome — hypokalaemia, hyperglycaemia, tremor, tachyarrhythmia and lactic acidosis — coherently and predictively, and nothing contradicts it, but the relative contributions of catecholamine release, adenosine antagonism and phosphodiesterase inhibition to any single sign have not been separated in people.
2. The adenosine-antagonism account of the seizures is on firmer ground Established and has real bedside consequences — it predicts both the low seizure threshold and the blunted benzodiazepine response the label warns of.² It is one of the better examples in the library of a receptor mechanism that earns its keep at the bedside.
3. Shannon's acute-versus-chronic distinction is the most important quantitative finding on this page, and it is robust — two prospective cohorts, 249 and 356 patients, reaching the same conclusion by the same route.³⁴ The specific risk ratios (4.85 for major toxicity; 4.97 for arrhythmia) come from single-centre poison-centre data and should be read as strong directional evidence, not precise population estimates.
4. Every EXTRIP recommendation rests on low-to-very-low quality evidence — 141 articles, 143 patients, 10 deaths — even though the dialysability grade is A.¹ The concentration cut-points are consensus positions built largely on case reports, and the age and chronicity thresholds are Shannon's findings translated into a rule rather than independently validated.
5. The therapeutic and toxic ranges quoted are the UK label's, and they are narrow by design.² The 20 mg/L ceiling and the >100 mg/L acute danger threshold are widely reproduced and consistent across sources; the specific half-life and protein-binding figures are formulation- and patient-dependent and are given here by the label's own values rather than as universal constants.
6. Theophylline is a declining problem, and that is itself a hazard. Shannon's 1999 conclusion — that outcomes had not improved across a decade and that the drug's indications should be re-examined with a view to minimal use⁴ — has largely come true, which means clinicians now see it rarely and may not recognise the chronic-toxicity presentation when it arrives in an older patient with an arrhythmia and an unremarkable level.

References

1. 1
   Ghannoum M, Wiegand TJ, Liu KD, Calello DP, Godin M, Lavergne V, Gosselin S, Nolin TD, Hoffman RS; EXTRIP Workgroup. Extracorporeal treatment for theophylline poisoning: systematic review and recommendations from the EXTRIP workgroup. Clin Toxicol (Phila) 2015;53(4):215–29. PubMed 25715736 · Recommendation set also published at extrip-workgroup.org/theophylline. Source of the level-A dialysability grading, every dialysis indication and threshold, the acute >100 mg/L and chronic >60/50 mg/L cut-points, the cessation threshold of <15 mg/L, the MDAC-during-dialysis recommendation, and the 141-article / 143-patient / 10-death evidence base. Verified 12 Sep 2026 from the abstract and the workgroup's published recommendation page.
2. 2
   Uniphyllin Continus 300 mg prolonged release tablets — Summary of Product Characteristics, Napp Pharmaceuticals. emc product 100851. §4.9 and §5.2 fetched and read in full. Source of the 20 mg/L therapeutic ceiling and do not exceed wording, the up-to-12-hour delayed-symptom warning for prolonged-release preparations, the ~100% bioavailability, ~60% protein binding, CYP-mediated metabolism to 1,3-dimethyluric acid and 3-methylxanthine, ~10% unchanged renal excretion and ~7 h half-life, the dose-proportional-only-across-200–600-mg statement, the transcellular-shift hypokalaemia with rebound-hyperkalaemia warning, the MDAC-enhances-elimination statement, the reduced-benzodiazepine-efficacy warning, the avoid-lidocaine warning, and the β-blocker-except-in-asthmatics note. The SmPC's stated fatal-dose figures were read and deliberately not reproduced on this page. Verified 12 Sep 2026.
3. 3
   Shannon M. Predictors of major toxicity after theophylline overdose. Ann Intern Med 1993;119(12):1161–7. PubMed 8239246 Prospective study of 249 patients. Source of the finding that chronic overmedication produced major toxicity in 49% versus 10% for acute intoxication (RR 4.85, 95% CI 2.96–7.94) despite lower peak concentrations (283 vs 777 µmol/L), and that peak concentration >100 mg/L predicts major toxicity in acute overdose while age >60 predicts it in chronic overmedication regardless of concentration. Verified 12 Sep 2026 from the abstract.
4. 4
   Shannon M. Life-threatening events after theophylline overdose: a 10-year prospective analysis. Arch Intern Med 1999;159(9):989–94. PubMed 10326941 Prospective cohort of 356 patients. Source of the 20.8% arrhythmia and 8.2% seizure rates, the 4.2% mortality with 73% of deaths occurring in chronic overmedication, the chronic-versus-acute arrhythmia rate (35% vs 10%, OR 4.97, 95% CI 2.68–9.23) and death rate (8% vs 2.5%, OR 3.20, 95% CI 1.01–10.39), and the conclusion that outcomes had not improved over the decade. Verified 12 Sep 2026 from the abstract.
5. 5
   TOXBASE — theophylline; aminophylline. National Poisons Information Service. toxbase.org (NHS login required. NPIS: 0344 892 0111. The authoritative UK source for treatment thresholds, charcoal regimens, potassium replacement and dialysis criteria. Login-gated, so not quoted here.)