Na+ blockers: phenytoin, carbamazepine, oxcarbazepine, lamotrigine, valproate(also other mechanisms). Ca2+(T-type) blockers: ethosuximide, valproate. GABA-potentiating: benzodiazepines, phenobarbitone, vigabatrin(irreversible GABA-transaminase inhibitor), tiagabine(GABA reuptake inhibitor). Glutamate modulators: felbamate(NMDA antagonist), perampanel(AMPA antagonist). SV2A modulator: levetiracetam. Multi/uncertain: valproate, topiramate, gabapentin/pregabalin(α2δ VGCC subunit — despite being designed as GABA analogues).
Generalized tonic-clonic: valproate, phenytoin, carbamazepine, lamotrigine. Absence: ETHOSUXIMIDE or valproate ONLY. TRAP: phenytoin/carbamazepine INEFFECTIVE, can WORSEN — absence = thalamocortical T-type Ca2+ oscillation, Na+ blockers don’t address it(can exacerbate circuit). Focal/partial: carbamazepine, phenytoin, lamotrigine, levetiracetam. Myoclonic: valproate, levetiracetam. Carbamazepine/phenytoin can WORSEN (same trap category). Status epilepticus: IV benzodiazepine(lorazepam/diazepam) FIRST → IV phenytoin/fosphenytoin or valproate for sustained control.
Phenytoin: USE-DEPENDENT Na+ block(binds INACTIVATED state preferentially) → greatest effect on rapidly-firing seizure-focus neurons, spares normal firing(same principle as local anaesthetics/class I antiarrhythmics). ZERO-ORDER kinetics at therapeutic-toxic range(hepatic enzymes saturate) → small dose ↑ near ceiling → disproportionate plasma level rise → narrow TI, careful titration/monitoring needed.
Carbamazepine: use-dependent Na+ blocker(same as phenytoin) + trigeminal neuralgia + bipolar mood stabilizer. POTENT enzyme INDUCER(own metabolism + co-drugs incl. OCPs↓efficacy). AUTOINDUCTION over first weeks → dose often needs ↑ shortly after starting.
Valproate: MULTI-mechanism(Na+ block + T-type Ca2+ block + ↑GABA synthesis/↓breakdown) → BROAD-spectrum(generalized+absence+myoclonic, rare drug covering this range). TERATOGENIC — NEURAL TUBE DEFECTS(disproportionately high risk vs alternatives) + ↓cognitive outcomes in utero → high-dose folic acid if used in childbearing potential, avoid if alternative exists. Also: hepatotoxic(rare, severe, young children+polytherapy), weight gain, tremor, reversible hair loss.
Ethosuximide: blocks T-type Ca2+ in THALAMIC neurons specifically — thalamocortical circuit generates 3Hz spike-wave via these channels → mechanism-MATCHED targeted therapy → narrow-spectrum(ineffective other seizure types, doesn’t address Na+/GABA mechanisms).
Lamotrigine: use-dependent Na+ blocker(phenytoin/carbamazepine family) but broader spectrum+favourable SE profile → PREFERRED in pregnancy among Na+-blocker group. Dose/titration-rate-dependent SEVERE RASH(SJS) → always start LOW, slow up-titrate; +valproate inhibits lamotrigine metabolism → ↑levels/rash risk → even more cautious titration when combined.
Levetiracetam: binds SV2A(synaptic vesicle protein) — DISTINCT mechanism from all above → clean interaction profile(minimal hepatic metabolism, few induction/inhibition interactions). Broad-spectrum. SE: behavioural/psychiatric(irritability, mood change, occasionally psychosis) — specific liability of otherwise well-tolerated drug.
Benzodiazepines/phenobarbitone: ↑GABA-A(same mechanism as Sedatives/Hypnotics topic). Used for ACUTE termination(IV lorazepam/diazepam, rapid onset), not chronic maintenance(tolerance develops with chronic benzo use). Phenobarbitone still 1st-line in NEONATAL seizures despite sedation/induction burden elsewhere.
Teratogenicity = class-wide concern, but valproate’s NTD risk disproportionately HIGH vs alternatives — differential risk, NOT blanket avoidance(uncontrolled seizures also carry fetal risk).
Enzyme INDUCERS(carbamazepine, phenytoin, phenobarbitone) vs INHIBITOR(valproate) → dense interaction web among AEDs + OCPs/warfarin/other AEDs — knowing inducer-vs-inhibitor predicts interaction DIRECTION without memorizing each pairing.
SJS/TEN: across several Na+-blockers(phenytoin, carbamazepine, lamotrigine) — genetic susceptibility(HLA-B*1502 + carbamazepine-SJS/TEN, esp. Han Chinese/SE Asian ancestry → pre-treatment genetic screening recommended where feasible).
Gingival hyperplasia = distinctive PHENYTOIN effect(mechanism unclear, altered folate metabolism+fibroblast proliferation theory) + coarsened facial features + hirsutism(long-term).
Absence-seizure trap(phenytoin/carbamazepine ineffective/worsening) = clearest illustration of organizing principle: drug choice must match the SPECIFIC ion-channel/circuit mechanism driving that seizure type, not reach for “strongest”/most familiar agent — Na+ blocker can’t help(can worsen) a T-type Ca2+-driven thalamocortical oscillation. Understanding MECHANISM, not memorizing a lookup table, prevents this real prescribing error.
This is the organizing clinical logic for the whole topic — drug choice depends on seizure type, not on ranking drugs by overall potency:
Phenytoin: blocks voltage-gated Na⁺ channels in a use-dependent manner — preferentially binds the inactivated state of the channel, so its blocking effect is greatest on neurons firing rapidly/repetitively (as in a seizure focus) while relatively sparing normal, lower-frequency neuronal firing — the same use-dependence principle already seen with local anaesthetics and class I antiarrhythmics, applied here to explain selective seizure-focus suppression. Zero-order kinetics at therapeutic-to-toxic doses (hepatic metabolizing enzymes become saturated within the clinical dosing range) is a specific, high-yield pharmacokinetic fact — small dose increases near the therapeutic ceiling can produce disproportionately large rises in plasma level, the mechanistic reason phenytoin has a narrow therapeutic index requiring careful, incremental dose titration and monitoring rather than the simple proportional dosing that first-order-eliminated drugs allow.
Carbamazepine: also a use-dependent Na⁺ channel blocker (same fundamental mechanism as phenytoin), additionally used for trigeminal neuralgia and as a mood stabilizer in bipolar disorder — a potent hepatic enzyme inducer (its own metabolism as well as that of numerous co-administered drugs, including oral contraceptives, reducing their efficacy — a frequently examined interaction), and a drug that induces its own metabolism (autoinduction) over the first few weeks of therapy, meaning the dose often needs upward adjustment shortly after starting to maintain the same plasma level, a specific, examined pharmacokinetic quirk.
Valproate (sodium valproate/valproic acid): genuinely multi-mechanism — Na⁺ channel blockade, T-type Ca²⁺ channel blockade, and enhancement of GABA synthesis/reduced GABA breakdown — this broad mechanism explains its correspondingly broad-spectrum efficacy across generalized, absence, and myoclonic seizure types (one of very few antiepileptics effective across this range). Teratogenic — specifically associated with neural tube defects (a distinctively high risk among antiepileptics, requiring high-dose folic acid supplementation if used in a woman of childbearing potential, and avoidance where a suitable alternative exists) and with reduced cognitive outcomes in children exposed in utero, making it a drug increasingly avoided in this population despite its broad efficacy elsewhere. Also hepatotoxic (rare but severe, particularly in young children on polytherapy) and causes weight gain, tremor, and reversible hair loss.
Ethosuximide: blocks T-type Ca²⁺ channels in thalamic neurons specifically — thalamocortical circuits generate the characteristic 3 Hz spike-and-wave discharge of absence seizures via these T-type channel-driven oscillations, so this is a genuinely targeted, mechanism-matched therapy rather than a broad-spectrum agent that happens to also work — narrow-spectrum for exactly this reason (ineffective for other seizure types, since it doesn’t address Na⁺-channel-driven or GABA-related seizure mechanisms).
Lamotrigine: use-dependent Na⁺ channel blockade (mechanistically similar family to phenytoin/carbamazepine) but with a broader spectrum of efficacy and a comparatively favourable side-effect profile, making it a preferred option in pregnancy among the Na⁺-channel-blocking group specifically — its most distinctive risk is a dose- and titration-rate-dependent severe skin rash (including Stevens-Johnson syndrome), which is why lamotrigine is always started at a low dose with slow up-titration, particularly when combined with valproate (which inhibits lamotrigine’s metabolism, raising levels and rash risk further, requiring an even more cautious titration schedule when co-prescribed).
Levetiracetam: binds synaptic vesicle protein 2A (SV2A), modulating neurotransmitter vesicle release — a mechanism genuinely distinct from every class above, contributing to its comparatively clean drug-interaction profile (minimal hepatic metabolism, few enzyme induction/inhibition interactions, an advantage in a patient already on multiple interacting drugs). Broad-spectrum efficacy, but behavioural/psychiatric adverse effects (irritability, mood change, occasionally psychosis) are a specific, examined liability of this otherwise well-tolerated drug.
Benzodiazepines/phenobarbitone: potentiate GABA-A receptor activity, mechanistically identical to their action described under Sedatives and Hypnotics — used for acute seizure termination (IV lorazepam/diazepam, given rapid onset) rather than chronic maintenance therapy in most cases (tolerance to the antiepileptic effect develops with chronic benzodiazepine use, and phenobarbitone’s sedation/enzyme-induction burden limits it to specific situations, notably still first-line in neonatal seizures).
The absence-seizure prescribing trap (phenytoin/carbamazepine ineffective or worsening) is the clearest illustration of this topic’s organizing principle: antiepileptic drug choice must match the specific ion-channel/circuit mechanism actually driving a given seizure type, not simply reach for “the strongest” or most familiar agent — a Na⁺ channel blocker cannot help a T-type Ca²⁺-channel-driven thalamocortical oscillation, and can worsen it by suppressing compensatory inhibitory circuitry, which is exactly why understanding mechanism rather than memorizing a drug-seizure lookup table is what actually prevents this specific, real prescribing error.
Personal revision notes, mnemonics and reminders.
