Introduction
Epilepsy is a serious neurological disorder that affects over 1% of the US population, and is associated with significant morbidity and mortality.  Quality of life can be a significant issue as well. Some deaths are related to the underlying condition that is responsible for the seizure disorder; however many are related to epilepsy, including deaths from seizures themselves.  Injury or death can also occur from drowning, vehicle wrecks or other types of accidents.  In addition, there is the major concern of Sudden Unexplained Deaths in Epilepsy (SUDEP).  Estimates of SUDEP death rates range from 0.35 to 1.2 per 1000 patient years in clinic based samples of adults with refractory partial epilepsy evaluated for epilepsy surgery.  Among those that continue to have intractable seizures following surgery the rate increases to 7.5 per 1000 patient years.  Given the significance of morbidity and mortality associated with uncontrolled epilepsy there is a definite need to optimally treat patients with refractory epilepsy[8].

Until 1991 only 6 major antiepileptic drugs (AEDs) were available for seizure control.  Each was associated with strong disadvantages including teratogenic potential, interaction with other drugs metabolized through the same pathways, strong sedative effects, and cognitive impairment.  With these AEDs seizure control was only initially achieved in half of the patients.  Additional dosage adjustments resulted in control for 20-30% leaving an additional 20-30% of patients remaining refractory to treatment[2,8].

Several physiological causes can be involved in epilepsy.  The excitatory amino acids, glutamate and aspartate, have been shown to be involved in the generation of seizures.  A seizure can be induced by systemic or intercerebral induction of these amino acids[1].   AEDs normally target one of several mechanisms  to help control seizure activity.  Among these are reduction of sodium and calcium conduction into the cell and modulation of the γ-Aminobutyric acid (GABA ) receptors or N-methyl-D-aspartic acid (NMDA) receptors.  NMDA receptors have also been shown to increase sodium and calcium influx into the cell, making them an excellent target for AEDs[1].

Felbamate (FBM), a broad-spectrum AED, was the first new major AED to be approved by the FDA since valproic acid was approved in 19938 followed by several others[1,2].  Felbamate is a dicarbamate that is chemically similar to Meprobamate, however it lacks the heavy sedating action8.  It was originally synthesized in the 1950’s but was not screened for antiepileptic properties until 30 years later[9,8].  FBM was approved as monotherapy as well as an adjunctive therapy for the treatment of partial and generalized seizures in adults as well as seizures associated with Lennox–Gastaut syndrome[8].

FBM was the first AED to be evaluated in a double blind study for patients undergoing evaluation for epilepsy, one of the first for be evaluated in a study as a monotherapy, and the first to be in a controlled clinical study for children with Lennox–Gastaut syndrome[2,8].  Ethical guidelines do not normally allow for drug testing on children. Unfortunately this leads to not knowing the effects of drugs on children prior to their release, but only gives a rough idea about efficacy.  Unexpected side effects relating to age specific organ toxicity, impact on behavior and cognition and developmental effects can remain unknown in children until after the medication is released[2].

After felbamate’s release several cases of aplastic anemia and liver failure were reported and warnings were sent out to medical professionals.  The FDA issued a black box warning and new evidence based guidelines for efficacy and tolerability were addressed[2].  The efficacy of the drug is such that under certain guidelines the medication is still used, though not as in as many cases and under much closer observation.  Between 3200-4200 new patients are prescribed felbamate annually, and over the past 10 years it is estimated that 35,000 new starts have occurred.  Therapies that carry high risks for new-onset patients may have a favorable risk/benefit ratio due to the significant mortality and morbidity issues associated uncontrolled seizures.   A strong understanding of the risk/benefit of the medication is necessary for both the physician and patient; however its use as an important therapeutic option for some patients with refractory epilepsy should not be overlooked[8].

Discovery
Felbamate or 2-phenyl-1,3-propanediol dicarbamate was first synthesized in the late 1950s while trying to explore the SAR of meprobamate, an anxiolytic (anti-anxiety) compound[9].  Its anti-epileptic properties were not known until  Wallace Laboratories submitted Felbamate to the Epilepsy Branch of the National Institute of Neurological Disorders and Stroke (NINDS) to be screened for AED properties some thirty years later[3,9].  Clinical trials were full of many firsts.  Felbamate was the first AED to be evaluated in a double blind study for patients undergoing evaluation for epilepsy, one of the first for be evaluated in a study as a monotherapy, and the first to be in a controlled clinical study for children with Lennox–Gastaut syndrome[2,8].  License was granted by the FDA in 1993[2,3,8,9] making it the first new AED in over 15 years[9].

Indications
Felbamate has demonstrated high efficacy in treating seizures that are refractory to many other AEDs including those associated with Lennox–Gastaut syndrome[1,2,8].  This is especially helpful since common AEDs fail in 20-30% of patients[1,2,8,9].  Felbamate received its black box warning due to the danger of aplastic anemia and liver failure; however with proper risk management, a risk/benefit ratio may still favor its use in many patients.  The FDA has approved its use as a monotherapy and as an adjunct medication to other AEDs.  It is mainly used to treat refractory partial seizures without generalization and Lennox–Gastaut syndrome1, but evidence also exists for use in pediatric treatment of infantile spasms, partial seizures, absence seizures, juvenile myoclonic epilepsy (JME), and Landau– Kleffner syndrome (LKS)[2].

A study of the long term efficacy in children was assessed during a 3 year follow up study of 36 pediatric patients that showed a progressive decline in seizure control over that period.  A reduction in seizure frequency in response to felbamate was 69% at 3 months, 47% at 1 year and 41% at 3 years.  The best results from this study were in patients with simple partial seizures as well as those with tonic and atonic seizures[3].

Another two studies in 1991 examined efficacy in children and adults when FBM was used as an adjunct to phenytoin or carbamazepine.   Small, but encouraging, improvements in seizure control were noted and later reports showed greater efficacy; 20% of adults and 53% of children with partial-onset seizures experienced at least a 50% reduction in seizure frequency while using FBM as an adjunct to a variety of AEDs[8].

Dosing
The recommended initial dosing in children is 15mg/kg/day to be titrated up to 45mg/kg/day over a period of three weeks[8].  Some children may require up to 90mg/kg/day for complete control[2] as clearance in children is faster.   In adults a 3600 mg/day is achieved in three weeks of titration.  Different serum levels are nessasary for different individuals.  European guidelines recommend lower starting doses and slower titration in an attempt to achieve better tolerability[8].  For those on a ketogenic diet the tablet form is recommended due to the substantial amount of sorbitol used in the liquid suspension[8].  Felbamate can also affect the serum levels of phenytoin, phenobarbital, valproic acid (VPA), and carbamazepine- 10,11-epoxide by up to 20–50%[2] so adjustments must be made if felbamate is used in adjunct with these medications.

ADME
FBM is well absorbed orally (~90%), has a volume of distribution of 0.81 liters/kg, and has less than a 25% binding to plasma protein[9], with peak plasma levels being achieved in 3-5 hours.  During clinical trials, linear pharmacokinetics were observed[8].  Half life of the medication in general is 14 to 23 hours but can be affected by adjunctive therapies[9].   FBM easily crosses the blood-brain barrier with a plasma to brain ratio of 0.69.

Metabolism occurs in the liver via the P450 system.  Isozymes CYP3A4 and CYP2E1 catalyze hydroxylated metabolites while felbamate inhibits CYP2C199.  About 20-30% of the polar metabolites remain unknown.  FBM undergoes extensive metabolism with a renal clearance of more than 95%[9].

Method of Action
While the exact methods of action of FBM are unknown, more information continues to surface.  Several  methods of action appear to be present1,5 including: (1) blockade of voltage-gated sodium channels[1.2,5,8]; (2) interference with both N-methyl-daspartate (NMDA) NR2B subunit[1,2,5,8] and non-NMDA subtypes of glutamate receptors; (3) inhibition of voltage-gated  calcium currents; (4) allosteric modulation of gammaaminobutyric acid (GABA) receptors[2,8]; (5) Enhances GABA-mediated events through a barbiturate-like modulatory effect on the GABAA receptor[1];  (6)reduction of glutamate-induced excitation[1,2,5];(7) Inhibits glutamatergic neurotransmission via AMPA/kainate receptors[1]. 

It is also found that use-dependant inhibition of NMDA currents occurs[5].  One reason for this is that FBM has been shown to accelerate the activation kinetics of NMDA receptors and slow the recovery from channel desensitization[5].This is strongly encouraging as one effect of seizure has been found in rat models is the increase of NMDA currents after seizure, causing a positive feedback that could be a factor leading to status eplepticus[10].

The relatively minor effect on GABA function my well point to the reason there is a lack of sedation and lack of effect on cognitive impairment that is seen in many AEDs[5,8].  One study on pediatric patients found that FBM was more effective on patients older than 10 due to the fact that clearance is higher in younger children[3].

Structure-Function
Many of the effects of FBM are not yet understood, but more research is ongoing to try to explain the effects of the drug.  Felbamate has been shown to block Voltage Gated sodium channels[6] as well as calcium channels.  This is only one small effect of the medication however.

FBM was demonstrated to bind to the NR1/NR2B NMDA receptor at two inner pairs of residues;  V644(NR1)-L643(NR2B)(the two inner pairs) and T648(NR1)-T647(NR2B) (the two outer pairs).  Removal of either of these sets of residues greatly decreases the inhibition cause by FBM, and removal of both pairs eliminates all effects of FBM on the NMDA receptor[5].  This points to the probability that these two sites are very likely the exclusive molecular determinants of the FBM binding site on NMDA[3].

FBM has also been shown to selectively bind to NMDA channels that are in an open state and in an especially desensitized state, but not in the closed state.  This is a plausible explanation for the use-dependant inhibition of NMDA receptors and may explain the lack of sedative effects.  It has also been demonstrated that FBM has a stronger (~10 fold) effect on NMDA receptors containing NR2B subunits over those containing NR2A5.  NR2A subunits are the primary type found in adults, whereas the NR2B are found more in the developing neurons of children.

At least one method found to explain the reduction of glutamate release is by blocking of presynaptic NMDA receptors.  This could reduce the tonic positive feedback that is facilitated by glutamate release[6].

The frequency, but not the amplitude, of excitatory postsynaptic currents in entorhinal cortex neurons (responsible for learning and memory) are reduced by FBM[6].  This may help to explain the modulation of neuronal processes without strong cognitive effects.

Side Effects
The most common adverse effects are anorexia, headache, insomnia, nausea, dizziness, and gait disturbance; these are reversible after discontinuation or dose reduction[2,8].  Temporary weight loss occurred in 57% of patients, anorexia in 20%, and insomnia was reported in 16% of patients[3].

FBM, as with most AEDs, must be titrated down and not abruptly stopped.  Immediate discontinuation can result in withdrawal seizures, including status epilepticus (a life threatening condition where the brain is in a state of persistent seizure.)  Withdrawal of FBM can also cause reductions in serum levels of phenytoin, valproate, and phenobarbital which may also increase chance of withdrawal seizures if FBM is used as an adjunct therapy.  Reductions of 300-600 mg/day at weekly intervals are reasonable in adults.  If faster removal is needed other medications for coverage, such as benzodiazepine, are suggested by some clinicians[8].

No serious side effects were found during initial drug trials and this sparked strong optimism; however, postmarking possible idiosyncratic adverse events were revealed when several cases of aplastic anemia and liver failure were reported.  34 cases of aplastic anemia have been reported to date, and 33 of those occurred in the first year after FBM approval by the FDA[8].

The estimated risk of aplastic anemia in FBM prescribed patients ranges from 1:20,000 patients[1] to 1:5000[2] with about half of the cases being fatal.  It is notable, however, that though a high degree of the condition may be eliminated by proper screening, since most patients with FBM induced bone marrow failure had either pre-existing blood dyscrasias or pre-existing immune disorders[1].  Patients less than 13 years of age have never had cases reported of FBM induced aplastic anemia.

Thirty-one cases of FBM related aplastic anemia were reported between January and October 1994.  Of those, only 23 met the full diagnostic criteria for aplastic anemia.  Eight cases had confounding diagnoses, and in 9 additional cases FBM was only one of the possible drugs involved.  Only 3 cases were found to be definitely related to FBM.  Of the 23 cases meeting the criteria, 7 patients died.  Results of the examination indicated an association between FBM and aplastic anemia.  The risk of aplastic anemia in the general populus is 2.0–6.0 per million per year.  In  patients who take felbamate the risk is estimated to be between 27 and 209 cases per one million with the most probable risk estimation at 127 cases per million.  For comparison, carbamazepine (another AED) carries an aplastic anemia risk between 5 and 20 cases per one million users[8].

Since the initial reports in 1994, (and changes in prescribing guidelines) the only case of aplastic anemia reported was in a 42 year old woman who had been on the medication for 8 years (reported in 2000).  She was treated with triple immunosuppressive therapy and Granulocyte colony-stimulating factor and fully recovered.  Also noteworthy in this case was that the patient had a melanoma removed a few years prior[8].

Risk factors that need to be considered in evaluation of felbamate-associated aplastic anemia include history of cytopenia, history of autoimmune disorder (such as  systemic lupus erythematosus, rheumatoid arthritis, Hashimoto’s thyroiditis, panhypogammaglobulinemia, idiopathic thrombocytopenia purpura), and a positive antinuclear antibody titer.  Of the patients who developed aplastic anemia while on felbamate 42% had a history of cytopenia prior to felbamate use, 52% had a history of allergy or significant toxicity to an AED, and 33% had evidence of an underlying autoimmune disease[8].

The first case of liver failure was reported in the first year of marketing.  A total of 18 cases of liver failure have been reported.  There were 9 cases in adults and 9 in children.  Felbamate monotherapy was only reported in one case[8].  The risk of fatal hepatotoxicity from FBM is estimated to range from 1:7000 to 1:22,000[2].  This is similar to older AEDs, and it does not significantly differ from that of valproate[1].  The mean time to presentation of symptoms was 217 days with a range of 25-939 days.  Of the 18 cases, 7 were ruled likely to be due to FBM and 9 cases were found to be unlikely.  Of the unlikely cases, 6 occurred in the context of status epilepticus, one patient had acetaminophen toxicity, one had hepatitis A and one had “liver shock”.  Five of the deaths were likely due to FBM[8].

Considering that 5 felbamate related liver failures occurred in 130,000 to 170,000 people, the estimated risk would be between 1:26,000 to 1:34,000 exposures.  In comparison valproate, another common AED, has a risk of death due to liver failure between 1:10,000 and 1:49,000 exposures[8].

Prior to initiating FBM therapy a complete risk assessment should be done, including documentation of informed consent.  In addition to considering the history as stated above, a hematologic evaluation should be performed.  A comprehensive metabolic and hematologic evaluation including LFTs and platelet counts that can provide a baseline measurement should be part of the pretreatment evaluation.  After initiation of treatment, monitoring through clinical examination for signs of toxicity, evaluation of seizure reduction, and frequent laboratory visits are necessary.  Since signs of aplastic anemia and liver failure are normally seen in the first 6-12 months of therapy, less frequent monitoring is permitted after several months of stability.  Many clinicians recommend biweekly blood monitoring for the first 3 months followed by every 6-12 months thereafter[8].

Patient and caregiver education is paramount regarding the early signs and symptoms of aplastic anemia and liver failure to prevent deaths in the rare case of these complications.  Thorough knowledge of early warning signs for self monitoring and having blood drawn at the first appearance of possible symptoms are vital to reducing risk of mortality.  Symptoms of aplastic anemia include increased bleeding, bruising, petechiae, susceptibility to infection, shortness of breath, fatigue, decreased alertness, shortened attention span, unusually pale skin color, dizziness, and lingering illness. Early hematological warning signs include declining reticulocyte counts and thrombocytopenia, which may be an indicator of an aplastic crisis. Felbamate should not be prescribed to a patient with a history of hepatic abnormality. Liver failure can occur without warning and be fulminant (extremely fast and severe onset), which makes monitoring for signs difficult. The first signs and symptoms of liver toxicity are generally lethargy, nausea, and vomiting. Felbamate should be discontinued if serum AST or ALT concentrations increase to two or more times the upper limit of normal or if symptoms are suggestive of liver failure[8].

Studies with FBM have been conducted that are consistent with the theory that compounds that cause Idiosyncratic Drug Reactions (IDR) undergo bioactivation to a highly reactive electrophilic metabolite that is capable of forming covalent protein adducts[9].  Possible screening tests for other risk factors are being developed that may help greatly in the future by finding reactive metabolites early in treatment or GSH depleation[9] in candidates for treatment.

Metabolites and IDR
Idiosyncratic Drug Reactions (IDR) in FBM related to aplastic anemia and liver failure are not known to relate to the pharmocophore of the drug.  It is believed instead to be related to a highly reactive electrophilic metabolite[9].

In 1994 when the reported toxicities were recognized the only four metabolites had been characterized were p-OHF, 2-OHF, MCF and CPPA.  Since a one-step four electron oxidation from the alcohol (MCF) to the acid (CPPA) is very unlikely, an intermediate aldehyde is suspected.  This aldehyde intermediate, 3-carbamoyl-2-phenylpropionaldehyde (aldehyde carbamate), could potentially undergo a β-elimination to form a β-unsaturated aldehyde, 2-phenylpropenal.  Β-unsaturated aldehydes are highly reactive electrophiles and very well may be responsible for the IDR in FBM[9].

A study was undertaken to support the hypothesis that atropaldehyde was the reactive metabolite associated with felbamate idiosyncratic toxicity.  Aldehyde carbamate and atropaldehyde were both synthesized and examined for cytotoxicity.  Growth inhibition of 50% (GI50) was used as a marker of cytotoxicity.  All other metabolites of FBM had GI50 values at least a 100 times greater than that of either atropaldehyde or aldehyde carbamate, showing the relative toxicity of these two metabolites[9].

Another important test was used to confirm the reactivity of atropaldehyde that utilized glutathione trapping.  Showing addition between a suspected reactive molecule and glutathione is a common test to show reactivity and toxicity for compounds.  Atropaldehyde was found to undergo facile addition with glutathione with a half-life of about 40 seconds, further demonstrating reactivity and electrophilicity of this FBM metabolite[9].

In vitro experiments with aldehyde carbamate showed reactivity and decomposed to form atropaldehyde and 4-hydroxy-5- phenyltetrahydro-1,3-oxazin-2-one (oxazolidine cyclization product).  At a pH of 7.4 the half-life of aldehyde carbamate was less them 30 seconds.  Under phosphate buffered conditions oxazolidine was the favored decomposition product; however, in longer incubations oxazolidine decomposed to form atropaldehyde.  With the fast kinetics for the formation of atropaldehyde it would be expected that atropaldehyde would react at the site of formation in the liver, which could account for hepatotoxicity.  This does not, however, account for the blood dysrcasia of aplastic anemia[9].

For atropaldehyde to travel to sites distal to the liver, it would have to be in a latent form.  It was hypothesized that oxazolidine may be this latent form due to its favorability in formation, and latent kinetics to form atropaldehyde.  Experiments have been preformed to specifically look at this possibility through a series of in vitro trapping experiments.  The half life of oxazolidine at pH 7.5 was found to be 0.4 hours, showing that it was stable enough to travel to distal sites from the liver.  Additional studies identified the oxazolidine in the urine of FBM patients, adding support to this hypothasis[9]. 

Glutathione should act as a protecting mechanism by binding up the reactive compound before it can cause damage.  Most FBM patients will maintain sufficient levels of glutathione to detoxify the atropaldehyde that results from metabolism of FBM.  Glutathione depletion may be the cause of observed toxicities in some patients.  Considering this possibility, a quantitative assay for metabolites in patient urine was developed and tested.  The assay focused on CPPA and atropaldehyde, which should be the ultimate fate of aldehyde carbamate.  In the average patient that would be able to safely metabolize FMB, the ratio of acid carbamate to mercapturates should be fairly constant.  Patients deviating from this norm may be at high risk for felbamate toxicity if; excretion of more mercapturates caused a relatively low ratio of CPPA to mercapturates or if GSH depletion produced fewer mercapturates causing a relatively high ratio of CPPA to mercapturates[9].

With this test the average ratio of CPPA to mercapturates was 2.02 with a standard deviation of 0.6 (99% confidence level).  One patient had a ratio of 16.7, greater than 20 standard deviations from the norm.  The urine sample was taken 72 hours after the cessation of FBM therapy due to the fact that the patient had developed neutropenia, a type of blood dysrcasia.  This is the first reported case where formation of a metabolite has been linked to clinical outcome[9].

Drug Interactions
Felbamate is primarily metabolized  by the hepatic cytochrome P450 system, and particularly the CYP3A4 enzyme.  In addition, FBM is an inhibitor of the enzyme CYP2C19.  Medications metabolized by these enzymes can have their steady state plasma concentrations affected.  Phenytoin, carbamazepine, and phenobarbital all increase the clearance of FBM by about 15% resulting in lower FBM serum levels.  When enzyme inducing drugs are removed they can increase FBM serum levels slightly.  This necessitates the examination of FBM levels to determine is a decrease in dose is warranted.  Additionally, serum phenytoin, being secondarily metabolized by CYP2C19, increases 30-50% when felbamate is added, and the effect of baseline felbamate will also be greater when the baseline of phenytoin is higher.  Phenobarbital levels increase 24% with the addition of FBM.  Carbamazepine, also metabolized by CYP3A4, decreases due to the addition of FBM, however, the degradation of the toxic metabolite carbamazepine epoxide is inhibited.  The β-oxidation of valproate is also inhibited, and can increase valproate serum levels 28-54%.  When felbamate is added as an adjunct to phenytoin, phenobarbital, carbamazepine, or valproate it is typically necessary to reduce their dosages by 25-33%[8].

Other drugs that are metabolized by the P450 system can also be affected.  Estrogen and progesterone oral contraceptives should have their levels monitored.  Felbamate has been shown to decrease AUCO-24 by 42%, though no effect has been observed in ethinyl estradiol.  No volunteer in the study showed hormonal evidence of ovulation during the study[8].

Even though FBM is very effective when used as an adjunctive therapy, monotherapy is preferred due to the possibilities of drug interactions and to improve tolerability.  If FBM is to be used as an adjunct it is preferred to use one of the AEDs that have not been shown to produce pharmacokinetic interactions[8].

Conclusions
Even with the development of new AEDs, 25-30% of children with epilepsy remain refractory.  Development of new AEDs should continue, especially targeting pediatric patients and the developing brain[2].

Felbamate has shown a very strong efficacy, coupled with extremely dangerous risks.  Epileptologists need to thoroughly appreciate the risk/benefit ratio of FBM to be able to evaluate its use for patients with refractory epilepsy.  Understanding risk factors can minimize occurrence of catastrophic events by greatly reducing the prescribing of this drug to high risk patients.  Since the first reports of idiosyncratic toxicities in 1994 there has only been one case of aplastic anemia.  During that time there have also been over 35,000 new starts on felbamate with no other reports of liver failure or aplastic anemia.  These facts and the strong efficacy of this drug in refractory epilepsy and looking at the dangers associated with refractory epilepsy seem to indicate that FBM has a risk/benefit ratio that could encourage its use in many cases of refractory epilepsy[8].

Sources
[1] Piotr Czapin´ski; Barbara Blaszczyk; Stanislaw J. Czuczwar.  Mechanism of Action of Antiepileptic Drugs. Current Topics in Medicinal Chemistry. 2005, 5, 3-14
[2] Hee Hwang; Ki Joong Kim. New antiepileptic drugs in pediatric epilepsy. Brain&Development, 2008, 30, 549–555
[3] Giangennaro Coppola. Treatment of Partial Seizures in Childhood. CNS Drugs, 2004, 18 (3), 133-156

[4] R. Scott Obach; Amit S. Kalgutkar; John R. Soglia; Sabrina X. Zhao.  Can In Vitro Metabolism-Dependent Covalent Binding Data in Liver Microsomes Distinguish Hepatotoxic from Nonhepatotoxic
Drugs? An Analysis of 18 Drugs with Consideration of Intrinsic Clearance and Daily Dose. Chem. Res. Toxicol. 2008, 21, 1814–1822

[5] Huai-Ren Chang; Chung-Chin Kuo. Molecular Determinants of the Anticonvulsant Felbamate Binding Site in the N-Methyl-D-Aspartate Receptor. J. Med. Chem. 2008, 51, 1534–1545

[6] Jian Yang; Caroline Wetterstrand; Roland S.G. Jones. Felbamate but not phenytoin or gabapentin reduces glutamate release by blocking presynaptic NMDA receptors in the entorhinal cortex. Epilepsy Research, 2007, 77, 157—164

[7] Jarogniew J. Luszczki; Neville Ratnaraj; Philip N. Patsalos; Stanislaw J. Czuczwar. Pharmacodynamic and Pharmacokinetic Interaction Studies of Loreclezole with Felbamate, Lamotrigine, Topiramate, and Oxcarbazepine in the Mouse Maximal Electroshock Seizure Model. Epilepsia, 2005, 46(3):344–355

[8] John M. Pellock; Edward Faught; Ilo E. Leppik; Shlomo Shinnar; Mary L. Zupanc. Felbamate: Consensus of current clinical experience. Epilepsy Research 2006, 71, 89–101

[9] Christine M. Dieckhaus; Charles D. Thompson; Shane G. Roller; Timothy L. Macdonald. Mechanisms of idiosyncratic drug reactions: the case of felbamate. Chemico-Biological Interactions 2002, 142,  99-117

[10] Russell M. Sanchez; Carl Wang; Ginger Gardner; Lianna Orlando; David L. Tauck; Paul A. Rosenberg; Elias Aizenman; and Frances E. Jensen. Novel Role for the NMDA Receptor Redox Modulatory Site in the Pathophysiology of Seizures. J. Neuroscience, 2000, 20(6), 2409–2417