N03: Antiepileptics

ATC N03 holds the antiepileptics, drugs that prevent seizures, grouped by chemical class, with many newer drugs in a catch-all group.
Updated

September 25, 2026

Antiepileptics, also called antiseizure drugs, prevent seizures, but there is no evidence that they modify the underlying disease [1].

How do antiepileptics work?

Antiepileptics damp down the excitability of the brain, by four broad mechanisms [1]:

  1. Ion channel modulation. Phenytoin (N03AB02), carbamazepine (N03AF01) and lamotrigine (N03AX09) act on voltage-gated sodium channels, ethosuximide (N03AD01) on calcium channels, and retigabine (N03AX21) on potassium channels.
  2. Enhanced GABA inhibition. Benzodiazepines, tiagabine (N03AG06) and vigabatrin (N03AG04) strengthen inhibitory neurotransmission by γ-aminobutyric acid (GABA).
  3. Reduced glutamate excitation. Perampanel (N03AX22) weakens excitatory neurotransmission by glutamate.
  4. Modulated neurotransmitter release. Levetiracetam (N03AX14) and brivaracetam (N03AX23) act presynaptically, as do gabapentin (N02BF01) and pregabalin (N02BF02).

Lacosamide (N03AX18) is a sodium channel drug of a different kind: instead of affecting fast inactivation, it selectively enhances the slow inactivation of the channel [2].

Drug interactions

Some patients with difficult-to-treat epilepsy benefit from a combination of two or more antiepileptics, and almost all patients take other drugs at some point. Carbamazepine, phenytoin, phenobarbital (N03AA02) and primidone (N03AA03) induce many cytochrome P450 and glucuronyl transferase enzymes. They can drastically reduce the serum concentrations of other drugs metabolized by these enzymes, lamotrigine among them. Valproic acid (N03AG01) induces no enzymes, but it inhibits the metabolism of some drugs, most notably phenobarbital and lamotrigine. Most newer antiepileptics are less likely to induce or inhibit these enzymes. Levetiracetam, gabapentin and pregabalin have not been reported to cause, or be the target of, clinically relevant pharmacokinetic interactions. Of the pharmacodynamic interactions, neurotoxic effects appear more likely when antiepileptics sharing the same primary mechanism of action are combined [3].

Efficacy

Efficacy is measured as the reduction in seizure frequency. For lacosamide, the relationship between plasma concentration and this reduction was described by an Emax model [2].

Seizure frequency is count data. A Poisson model assumes that the variance equals the mean and that counts in separate intervals are independent. Daily seizure counts from 551 patients with treatment-resistant (refractory) partial seizures broke both assumptions. They varied more than a Poisson model predicts (overdispersion), and a negative binomial model described them best. Letting the parameters depend on whether the previous day had a seizure, a Markov element, improved the fit further [4].

In an exposure-response analysis of add-on pregabalin in 1042 patients, monthly seizure frequency was modeled as a Poisson process with a baseline, a placebo effect and a drug effect. The model split the patients into two subpopulations: 75% had a dose-related decrease in seizure frequency, and 25% had no significant decrease. Among the responders, 186 mg per day gave half of the maximal reduction [5].

Safety

The older antiepileptics phenytoin, carbamazepine and valproic acid have pronounced pharmacokinetic variability between patients and a narrow therapeutic range. Their dose has commonly been adjusted to reach a serum concentration within a predefined therapeutic range. These ranges must be interpreted with caution, since many patients do best at concentrations below or above them [6]. For phenytoin, part of the variability comes from variants in the CYP2C9 gene, and the HLA-B*15:02 allele raises the risk of severe skin reactions (Stevens-Johnson syndrome and toxic epidermal necrolysis) [7].

Antiepileptics differ in their risk of causing birth defects (teratogenic risk), which also depends on the dose. In the EURAP pregnancy registry, major birth defects (congenital malformations) followed 10.3% of pregnancies exposed to valproic acid alone, against 2.8% to 3.0% for levetiracetam, lamotrigine and oxcarbazepine (N03AF02). The risk rose with the dose at conception for carbamazepine, lamotrigine, phenobarbital and valproic acid [8].

References

[1]
Sills GJ, Rogawski MA. Mechanisms of action of currently used antiseizure drugs. Neuropharmacology 2020;168:107966. https://doi.org/10.1016/j.neuropharm.2020.107966.
[2]
Cawello W. Clinical pharmacokinetic and pharmacodynamic profile of lacosamide. Clinical Pharmacokinetics 2015;54:901–14. https://doi.org/10.1007/s40262-015-0276-0.
[3]
Perucca E. Clinically relevant drug interactions with antiepileptic drugs. British Journal of Clinical Pharmacology 2006;61:246–55. https://doi.org/10.1111/j.1365-2125.2005.02529.x.
[4]
Trocóniz IF, Plan EL, Miller R, Karlsson MO. Modelling overdispersion and Markovian features in count data. Journal of Pharmacokinetics and Pharmacodynamics 2009;36:461–77. https://doi.org/10.1007/s10928-009-9131-y.
[5]
Miller R, Frame B, Corrigan B, Burger P, Bockbrader H, Garofalo E, et al. Exposure-response analysis of pregabalin add-on treatment of patients with refractory partial seizures. Clinical Pharmacology & Therapeutics 2003;73:491–505. https://doi.org/10.1016/S0009-9236(03)00049-3.
[6]
Johannessen SI, Tomson T. Pharmacokinetic variability of newer antiepileptic drugs. Clinical Pharmacokinetics 2006;45:1061–75. https://doi.org/10.2165/00003088-200645110-00002.
[7]
Karnes JH, Rettie AE, Somogyi AA, Huddart R, Fohner AE, Formea CM, et al. Clinical pharmacogenetics implementation consortium (CPIC) guideline for CYP2C9 and HLA-b genotypes and phenytoin dosing: 2020 update. Clinical Pharmacology & Therapeutics 2021;109:302–9. https://doi.org/10.1002/cpt.2008.
[8]
Tomson T, Battino D, Bonizzoni E, Craig J, Lindhout D, Perucca E, et al. Comparative risk of major congenital malformations with eight different antiepileptic drugs: A prospective cohort study of the EURAP registry. The Lancet Neurology 2018;17:530–8. https://doi.org/10.1016/S1474-4422(18)30107-8.