How did phenobarbital’s chemical structure affect the development of subsequent antiepileptic drugs (AEDs)?


  • Meir Bialer

    1. School of Pharmacy, Institute for Drug Research, Faculty of Medicine, and David R. Bloom Center for Pharmacy, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem, Israel
    Search for more papers by this author

Address correspondence to Meir Bialer, School of Pharmacy, Institute for Drug Research, Faculty of Medicine, The Hebrew University of Jerusalem, POB 12065, Jerusalem 91120, Israel. E-mail:


Phenobarbital has been in clinical use as an antiepileptic drug (AED) since 1912. The initial clinical success of phenobarbital and other barbiturates affected the design of subsequent AEDs (e.g., phenytoin, primidone, ethosuximide), developed between 1938 and 1962, the chemical structures of which resemble that of phenobarbital. However, the empirical discovery of carbamazepine (1962) and the serendipitous discovery of valproic acid (1967) led to subsequent AEDs having chemical structures that are diverse and completely different from that of phenobarbital. Sixteen AEDs were introduced between 1990 and 2012. Most of these AEDs were developed empirically, using mechanism-unbiased anticonvulsant animal models. The empirical nature of the discovery of these AEDs, coupled with their multiple mechanisms of action, explains their diverse chemical structures. The antiepileptic market is therefore crowded. Future design of new AEDs must have a potential for treating nonepileptic central nervous system (CNS) disorders (e.g., bipolar disorder, neuropathic pain, migraine prophylaxis, or restless legs syndrome). The barbiturates were once used as sedative-hypnotic drugs, but have been largely replaced in this role by the much safer benzodiazepines. In contrast, phenobarbital is still used worldwide in epilepsy. Nevertheless, the development of nonsedating phenobarbital derivatives will answer a clinical unmet need and might make this old AED more attractive.

The following sixteen new antiepileptic drugs (AEDs) were approved between 1990 and 2012 by the U.S. Food and Drug Administration (FDA) and/or by the European Medicines Agency (EMA): eslicarbazepine acetate (ESL), felbamate (FBM), gabapentin (GBP), lacosamide (LCS), lamotrigine (LTG), levetiracetam (LEV), oxcarbazepine (OXC), perampanel (PER), pregabalin (PGB), retigabine (RTG) or ezogabine, rufinamide (RUF), stiripentol (STR), tiagabine (TGB), topiramate (TPM), vigabatrin (VGB), and zonisamide (ZNS). ESL and STR are currently approved only in Europe. These AEDs offer appreciable advantages in terms of their favorable pharmacokinetics, improved tolerability, and lower potential for drug interactions. In addition, the availability of old and new AEDs with various activity spectra and different tolerability profiles enables clinicians to better choose an AED according to the characteristics of individual patients (Perucca et al., 2007; Bialer & White, 2010).

The AEDs approved between 1990 and 2012 divide into two categories (Backonja et al., 2011; Bialer, 2006, 2012; Bialer & White, 2010):

  • 1 Completely new chemical structures (e.g., LCS, RTG, or PER);
  • 2 Derivatives of existing AEDs that are second-generation or follow-up compounds of established AEDs, such as:
    •   a. OXC and ESL, which are carbamazepine (CBZ) derivatives;
    •   b. LEV, an ethyl derivative of the cognitive enhancer piracetam;
    •   c. PGB, a follow-up compound to GBP;
    •   d. GBP enacarbil, a recently introduced GBP prodrug, approved by the FDA for treating restless legs syndrome (2011) and postherpetic neuralgia (2012).

Although the new AEDs with a completely new chemical structure were developed empirically and in many cases serendipitously, the purpose of their design was either to widen the CNS activity or to improve efficacy, safety, and/or tolerability. The incentives for the design and development of second-generation AEDs were the following:

  • 1 Enhancement of brain penetration compared to the parent compound (e.g., PGB or GBP enacarbil compared to GBP);
  • 2 Elimination of parent compound toxic metabolite (e.g., VPA);
  • 3 Circumvention of the structural requirements for the teratogenicity of the parent compound (e.g., VPA) (Bialer, 2006; Bialer & Yagen, 2007).

This article focuses on the chemical properties of the first established AED, the barbiturate phenobarbital, and discusses how its chemical structure has affected the design and development of subsequent new AEDs from both categories (completely new chemical structures and second-generation) (Bialer, 2006, 2012; Bialer & Yagen, 2007; Bialer & White, 2010). In addition to aspirin (1899) and paracetamol or acetaminophen (first used in medicine in 1893 but only gained popularity after 1949) (Grosser et al., 2011), phenobarbital is the only synthetic drug and the only barbiturate that is still widely used worldwide 100 years after its introduction to the clinical practice.


Phenobarbital, 5-ethyl-5-phenyl barbituric acid, is the second barbiturate in clinical use, introduced for the treatment of epilepsy in 1912 following barbital (Veronal, or 5,5-diethyl barbituric) that was introduced in 1903 (Cozanitis, 2004).

Adolf von Baeyer reported the discovery of barbituric acid or malonylurea in 1864 and Mulder subsequently confirmed its chemical structure in 1873 (Vida & Gerry, 1977). There are three hypotheses about the origin of the term “barbiturates” (Dundee & Mellroy, 1982; Lopez-Munos et al., 2005):

  • 1 von Baeyer is said to have used this name for sentimental reasons, in honor of his friend Barbara (Cohen, 1943);
  • 2 von Baeyer’s discovery was on December 4, the feast day of Saint Barbara, patron saint of artillery men, tunnel diggers, and firemen. Therefore von Baeyer designated his malonylurea barbituric acid (Sharpless, 1970; Cozanitis, 2004);
  • 3 The term was inspired by the “barbed” appearance of the crystals of these ureic compounds or uredines (Fieser, 1944). In any case, it is clear that the combination of the terms “barb(ara)” and “urea” forms the basis for the name barbiturates (Lopez-Munos et al., 2005).

However, barbituric acid per se does not possess CNS activity, probably due to its lower lipophilicity compared to its alkyl or aryl derivatives, namely the barbiturates. Fischer and von Mering (1903) employed a condensation reaction to synthesize barbituric acid, using diethyl malonic acid and urea; this was not only a practical method for synthesizing barbiturates but was also coupled with the revolutionary discovery that barbital possesses hypnotic properties. Having visited Verona earlier, von Mering thought it appropriate to name the new compound after this city; thus barbital became Veronal. Another theory is that the name Veronal (from the Latin versus, meaning true) was coined by Fischer, who claimed to have found the “true” hypnotic compound (Sneader, 1985). Veronal was manufactured by E. Merck (Germany) and by Winthrop in the United States. Thus began the barbiturates era (Table 1) (Cozanitis, 2004). In his memoirs, the Nobel laureate Emil Fisher mentioned Veronal only in passing, since it was the synthesis of enzymes that most interested him (Cozanitis, 2004).

Table 1.   Chemical structures, pharmacokinetics, and pharmacodynamic properties of barbiturates that are currently available commercially (asterisks denote chiral centers, partly based on Mihic & Haris, 2011 ) Thumbnail image of

Barbiturates were introduced into clinical practice during the first decade of the 20th century. Their history as AEDs began with a report from Hauptmann (1912) that a patient with epilepsy had fewer seizures when given phenobarbital for sedation. In 1919, Horlein at Bayer introduced phenobarbital (Luminal), and the patent rights were granted to Bayer in 1916. Phenobarbital was rapidly recognized as a better and safer AED than bromides; consequently it replaced bromides, which had been the only AEDs used since 1857. Phenobarbital had a more prolonged pharmacologic action than its predecessor and soon became the “king of the barbiturates,” both in the hospital and in outpatient care (Shorter, 1997).

Veronal and phenobarbital were the first barbiturates to be accepted by the international pharmacopoeia. Acceptance in Britain and the United States was in 1914 and 1923, respectively. Phenobarbital is considered as a sedative rather than hypnotic substance and, according to its Controlled Substances Act grouping, it has low abuse potential. Tolerance in patients treated with phenobarbital has been reported but the addiction to phenobarbital was questionable (Cozanitis, 2004).

Phenobarbital is a weak acid (pKa = 7.3) that is sparingly soluble in water (1 mg/ml). Phenobarbital sodium salt (PB-Na) has a better water solubility than phenobarbital per se (free acid) and consequently, has been used in phenobarbital parenteral preparations. Nevertheless, a phenobarbital injection is not an aqueous solution but contains 20% PB-Na in a mixture of 90% propylene glycol and 10% water at pH = 10–11 (Bialer, 2009). Phenobarbital’s pKa (−log of the dissociation constant − Ka) is similar to that of the physiologic blood pH, and therefore phenobarbital is 50% ionized and 50% nonionized at pH = 7.3, but the ratio of ionized to nonionized phenobarbital changes according to the physiologic pH (Anderson & Levy, 1995).

Phenobarbital has a complete oral bioavailability (F = 100 ± 11%), very low total clearance (CL = 0.06 ± 0.01 ml/min/kg) that is 75% (hepatic) metabolic and 25% renal, a volume of distribution less than the total body water (V = 0.54 ± 0.03 L/kg), and a long half-life (t1/2 = 99 ± 18 h) (Browne et al., 1985; Anderson, 2002). Phenobarbital is mainly metabolized to two inactive primary metabolites: p-hydroxy-PB that is excreted in the urine as free and glucuronide conjugate, and an N-glucoside conjugate of phenobarbital (Anderson, 2002).

Sandberg (1951) postulated that in order to possess good hypnotic activity, barbituric acid must satisfy two criteria: (1) it must be a weak acid, and (2) it must have a lipid/water partition coefficient within a certain limit. Subsequently, the barbiturates were classified into potentially active and inactive classes. In the active class (as hypnotics) there were 5,5-disubstituted barbituric acids and thiobarbituric acids as well as 1,5,5-trisubstituted barbituric acids (Sandberg, 1951; Vida & Gerry, 1977). Substitution of various alkyl and/or aryl moieties for the N and C-5 hydrogens of barbituric acid yielded molecules that varied considerably in their dominant pharmacologic effect on laboratory animals, including inactive, sedative, and proconvulsant compounds (Prichard, 1980).

In 1922, Dox synthesized butabarbital (Table 1), a butyl analog of Veronal that was three times stronger with a shorter duration of action and thus lower “rebound” possibility. In 1923, Shonle and Molen synthesized amobarbital (Amytal) that was methyl homolog of butabarbital. Six years later, Amytal sodium became the first ever barbiturate to be used as an intravenous anesthetic (Cozanitis, 2004; Lopez-Munos et al., 2005).

The ease of substituting various moieties on position 5 of the barbituric acid molecule yielded over 2,500 barbiturates. About 50 barbiturates were marketed, and classified as short-, medium-, or long-acting sedatives (Cozanitis, 2004; Lopez-Munos et al., 2005). Although the barbiturates were used extensively in the first half of the 20th century as sedatives, hypnotics, and anesthetic drugs, today they have been largely replaced in these roles by the safer benzodiazepines, except for a few specialized uses. Currently, about eight barbiturates (Table 1) are sufficient to cover the therapeutic applications (e.g., insomnia, anesthesia) that still require barbiturates (Lopez-Munos et al., 2005). In contrast to other barbiturates, phenobarbital is still the most widely used AED in the developing world and remains a popular AED in many developed countries (Kwan & Broide, 2004). Phenobarbital’s centenary will be celebrated in 2012 with a special Centenary Symposium during the 10th European Congress on Epileptology (ECE) in London (September 30–October 4, 2012).

Other Phenobarbital Derivatives Used as AEDs


Phenobarbital’s success led to the development of other barbiturates as subsequent AEDs, including the N-methyl barbituric acid derivatives (Fig. 1): mephobarbital (N-methylphenobarbital), introduced in 1932 (Blum, 1932) and metharbital (5,5-diethyl-1-methylbarbituric acid), which was introduced in 1948 but never became popular (Eadie & Hooper, 1995, 2002). Like phenobarbital, both mephobarbital and metharbital are water-insoluble weak acids with pKa values of 7.8 and 8.5, respectively. The introduction of the N-methyl group into the phenobarbital molecule breaks the symmetric axis possessed by barbituric acid or phenobarbital. Consequently, unlike phenobarbital, mephobarbital is a chiral compound containing one asymmetric carbon atom at position 5 of the molecule. It has been used clinically as racemic mixtures [equal parts of (R)- and (S)-enantiomers]. Lim and Hooper (1989) and Hooper and Qing (1990) showed that mephobarbital metabolism is stereoselective with (R)-methylphenobarbital being metabolized by cytochrome P450 (CYP)2C19-mediated aromatic hydroxylation (a genetic polymorphism-susceptible metabolic pathway coregulated by mephenytoin hydroxylation), whereas (S)-methylphenobarbital undergoes CYP2D6-mediated demethylation to form phenobarbital (Lim & Hooper, 1989; Hooper & Qing, 1990; Eadie & Hooper, 2002). In contrast to phenobarbital, mephobarbital is currently not widely used.

Figure 1.

Chemical structures of phenobarbital and its subsequent derivatives and related AEDs. (Taken in part from Bialer, 2012).


In 1952, primidone (PRM, Fig. 1), was introduced as a new AED. PRM is a water-insoluble deoxyphenobarbital that differs from phenobarbital in its lack of the carbonyl group at position 2 of the pyrimidine ring (Bogue & Carrington, 1953; Bourgeois, 1995). Initially, PRM aroused great therapeutic interest, as it was considered more potent than phenobarbital and other barbiturates, but without sedative effects (Bogue & Carrington, 1953); this interest soon waned after it was discovered that phenobarbital is a metabolite of PRM (Buttler & Waddell, 1956). Indeed PRM is biotransformed to two active primary metabolites, phenylethylmalonamide (PEMA) and phenobarbital, which have longer half-lives than PRM. The fraction of PRM metabolized to phenobarbital is increased when PRM is concomitantly administered with enzyme-inducing AEDs. Therefore, the clinical use of PRM, even as monotherapy, involves three active entities, the plasma concentration ratio of which is variable. This has led to confusion about whether PRM’s antiepileptic activity results from the parent compound or its active metabolites, particularly since the PB-to-PRM brain ratio is 2 (Bourgeois et al., 1983a,b). After 60 years of clinical experience, PRM is not considered superior to phenobarbital, and hence its current clinical use is limited compared to phenobarbital.


Eterobarb, or N,N′-dimethoxymethylphenobarbital (Fig. 2), is a dimethoxymethyl derivative of phenobarbital with similar anticonvulsant properties and attenuated hypnotic activity compared to phenobarbital (Gallagher, 1986; Wolter, 1991). In humans, following oral dosing, eterobarb is a prodrug to phenobarbital; following intravenous administration, it has a very short half-life of 10–100 min. Eterobarb (Antilon), developed in the early 1980s by Ciba-Geigy, reached phase II clinical trials in patients with epilepsy (Wolter, 1991). However, its development was stopped presumably because, as a phenobarbital prodrug, it offered no major clinical advantages over the parent compound.

Figure 2.

The barbiturate (i.e., butabarbital) chemical origin of two CNS-active derivatives of valproic acid: valnoctamide (VCD) and valnoctyl urea (Asterisks denote chiral centers; Shimshoni et al., 2007).

T2000 and T2007

5,5-Diphenylbarbituric acid (DPBA) was synthesized in 1935 because of its structural similarity to phenobarbital and PHT, and it was evaluated as a potential hypnotic in rats (McElvian, 1935). Raines et al. (1973) found that DPBA possesses anticonvulsant activity. Following intraperitoneal administration to mice, its median effective dose (ED50) values were 63 and 26 mg/kg, respectively, in the maximal electroshock (MES) and subcutaneous Metrazol (scMet) seizure tests (Bialer et al., 2009). Following oral administration to rats, DPBA had ED50 values of 11.5 and 130 mg/kg in MES and scMet tests, respectively (Raines et al., 1975). Because of its anticonvulsant activity and lower hypnotic effect, and since DPBA could not be patented, Taro Pharmaceuticals developed a new DPBA prodrug named T2000 (Fig. 1) in the 1990s (Bialer et al., 2009). T2000, 1,3-dimethoxymethyl-DPBA (Fig. 1) was developed for the treatment of essential tremor, myoclonus–dystonia, and epilepsy (Bialer et al., 2009). Following oral dosing to humans, T2000 is biotransformed to two primary metabolites, DPBA and monomethoxymethyl-DPBA, whose terminal half-lives are 27–65 and 8–27 h, respectively. Another DPBA derivative that was developed is T2007, a sodium salt of DPBA (Fig. 1). T2000 and T2007 have the same active entity (DPBA) (Bialer et al., 2009, 2010). The current status of T2000 and T2007 is unclear since Taro Pharmaceuticals was recently acquired by Sun Pharmaceuticals. Nevertheless, development of a nonsedating phenobarbital derivative (e.g., T2000) will answer a clinical unmet need and might make this old AED more attractive.

Phenobarbital-Related Compounds

Phenytoin, mephenytoin, and fosphenytoin

Blitz had already synthesized phenytoin (PHT) or 5,5-diphenylhydantoin in 1908, 30 years before Merritt and Putnam’s discovery of PHT’s anticonvulsant efficacy (Kupferberg, 1995). In 1914, 5-ethyl-5-phenyl-hydantoin (Nirvanol) was used clinically as a sedative and anticonvulsant. Nirvanol is a hydantoin derivative with a chemical structure analogous to that of phenobarbital. However, Nirvanol’s clinical use declined due to a high frequency of skin rashes and fever (Kupferberg, 1995). PHT became available as an AED in 1938 after the discovery of its antiepileptic efficacy by Merritt & Putnam (1938). PHT’s discovery led to the findings that in the search for new AEDs:

  • 1 The efficacy of active anticonvulsant compounds (e.g., PHT), hitherto assessed solely by clinical observation, was now testable in preclinical studies that led to anticonvulsant animal models;
  • 2 To be effective, an AED does not need to be a sedative (Porter, 1986). It is still unclear whether PHT’s lack of sedative properties (compared to phenobarbital) is due to the hydantoin ring or to the different alkyl/aryl side-chain moieties (diphenyl compared to phenyl and ethyl in the case of phenobarbital).
  • 3 Vida and Gerry summarized the following structure activity relationship (SAR) requirements for anticonvulsant activity of hydration derivatives (Vida & Gerry, 1977; Kupferberg, 1995):
  •   a. Substitution in position 5 of the hydantoin molecule was critical, with a phenyl moiety being a preferred substitution;
  •   b. Two phenyl moieties at position 5 led to maximal anticonvulsant activity;
  •   c. Substitution of a phenyl with an alkyl group broadened the anticonvulsant spectrum;
  •   d. Substitution of a methyl or an ethyl group in position 3 of the hydantoin molecule increased the anti-scMet activity. Removal of the phenyl group or substitution with larger aryl groups (e.g., benzyl) decreases or eliminates the anticonvulsant activity (Vida & Gerry, 1977; Kupferberg, 1995).

Mephenytoin (3-methyl-5-ethyl-5-phenylhydantoin) has been marketed since 1945 and was initially thought to be more effective than PHT (Clein, 1945). However, in mice and rats it was less potent than PHT at the MES test and in rats it was equipotent to PHT. Mephenytoin is effective in treating partial-onset and secondarily generalized seizures; however, its clinical use has been limited due to its association with idiosyncratic side effects such as rash, fever, generalized lymphadenopathy, and fatal blood dyscrasias (Kupferberg, 1995).

Mephenytoin has one chiral carbon at position 5 (Fig. 1) and is available commercially as a racemic mixture. Mephenytoin’s pharmacokinetics is enantioselective with (R)-mephenytoin, undergoing CYP3A4-mediated N-demethylation to Nirvanol and (S)-mephenytoin undergoing CYP2C19-mediated para-hydroxylation that is susceptible to genetic polymorphism (Kupfer et al., 1980, 1981).

Fosphenytoin (FOS)

Fosphenytoin (FOS) is a sodium phosphate ester of 3-hydroxymethyl-5,5-diphenylhydantoin (Fig. 1), which is a parenteral water-soluble prodrug of PHT. The water solubility of FOS is 75,000 mg/L, whereas the water solubility of phenytoin sodium (PHT-Na, the parent compound in PHT injection) is only 20 mg/L. Consequently, the vehicle of FOS injection is a pure aqueous solution (pH adjusted to 8.6–9.0), whereas the vehicle in PHT-Na parenteral preparation is 40% propylene glycol, 10% alcohol, and 50% water (pH adjusted to 12). Therefore the injection cannot be diluted to form an infusion. Administration of PHT-Na intravenously is painful and can cause vascular tissue damage and phlebitis. Local and systemic adverse effects of PHT-Na injection have been attributed to the relatively high pH of PHT-Na injection’s vehicle (Jamerson, 1994). Following parenteral administration, FOS is biotransformed rapidly to PHT, with a half-life of 8–15 min and clearance of 19.8 ± 1.6 L/h, and appears to be better tolerated than PHT-Na (Leppik et al., 1990; Pryor & Ramsay, 2002).


Ethosuximide and methsuximide

The two anticonvulsant succinimides currently in clinical use whose chemical structures resemble that of PB are ethosuximide and, to a lesser extent, methsuximide (Fig. 1). Ethosuximide evolved from an SAR study initiated by Miller & Long (1953) aimed to design and develop a drug that was effective against absence seizures with minimal adverse reactions. Succinimide per se has no anticonvulsant activity, but the introduction of ethyl and methyl moieties at position C-2 of the succinimide molecule resulted in ethosuximide, which was remarkably effective in controlling absence seizures (Sherwin, 1978). The introduction of a single phenyl moiety in C-2 of the succinimide molecule resulted in methsuximide, which was less effective than PHT. Vida and Gerry concluded that for a succinimide derivative to possess anticonvulsant activity, it has to have two substitutions at the C-2 position of the molecule (e.g., ethosuximide) or a substitution at C-2 and on the nitrogen (e.g., methsuximide). A small dialkyl substitution at C-2 leads only to activity in the scMet rodent model indicative of anti-absence activity. Anticonvulsant activity in both the MES and scMet tests is achieved with a phenyl substitution at C-2 and an alkyl on the nitrogen of the succinimide molecule. 2,2-Diphenyl-succinimide resembles PHT and thus has mainly anti-MES activity and is more active than its constitutional isomer 2,3-diphenyl-succinimide (Vida & Gerry, 1977; Ferrendelli & Kupferberg, 1980).

The discovery of ethosuximide was also motivated by a search in the 1950s for new antiabsence drugs that would be more effective and less toxic than trimethadione (introduced in the 1940s) for treating absence or myoclonic seizures.

Carbamazepine (CBZ)

Carbamazepine (CBZ) is an iminostilbene derivative developed in the late 1950s and early 1960s that is chemically related to the tricyclic antidepressants (e.g., imipramine, desipramine). CBZ differs from imipramine only by its double bond between C-10 and C-11 and a shorter side chain. In contrast, the chemical structures of CBZ and its follow-up AED, OXC (Bialer, 2002) and ESL (Bialer & Soares-da-Silva, 2012), are different from that of phenobarbital and all other AEDs that preceded CBZ. Unlike many other AEDs, CBZ lacks a saturated carbon atom, has an amide moiety that is not part of a heterocyclic ring, and has a tricyclic structure (Kutt, 1978; Suria & King Killam, 1980). Despite their different chemical structures, the anticonvulsant profile of CBZ is generally similar to that of PHT, indicating that the mechanism of action (MOA) of these two old and established AEDs is not yet completely understood.

Valproic Acid (VPA)

Valproic acid (VPA, di-n-propylacetic acid) was first synthesized in 1882 by Burton (1882), but had no known clinical use until Eymard serendipitously discovered its anticonvulsant activity in Carraz’s laboratory in 1962 (Meunier et al., 1963; Loescher, 1999). VPA’s first clinical trial was published in 1964 (Carraz et al., 1964); in 1967 it was approved for marketing in France, and subsequently it was marketed worldwide.

The serendipitous discovery of VPA explains why its chemical structure is so different from that of phenobarbital and all other AEDs that preceded VPA. VPA is an achiral short-branched fatty acid with eight carbons that does not contain a nitrogen atom or a cyclic ring in its chemical structure. Therefore, VPA is a water-insoluble (water solubility is 1.27 mg/ml) weak acid (pKa = 4.8) whose sodium salt is freely water soluble and hygroscopic (Bialer, 2012). Chemically, VPA is one of the simplest drugs currently available in our therapeutic arsenal that is also approved (by FDA and EMA) for migraine prophylaxis and for bipolar disorder. Even after 45 years of clinical experience and despite its simple chemical structure, VPA has the potential for new roles beyond epilepsy (Nalivaeva et al., 2009).

Follow-up AEDs to Valproic Acid (VPA)

In anticonvulsant animal models, VPA is the least potent of the established AEDs; this is reflected in VPA’s high dose and therapeutic plasma levels (Bialer et al., 2004). In addition, VPA’s clinical use is limited by two rare but serious side effects: teratogenicity and hepatotoxicity. These side effects restrict its use in women of childbearing age and in children.

Two VPA amide derivatives have been in clinical use in Europe. Valpromide (VPD)—VPA’s corresponding amide—is still used to treat epilepsy and psychiatric disorders. VPD’s constitutional isomer, valnoctamide (VCD), was in clinical use (as a racemate) as an anxiolytic between 1964 and 2005, but its marketing was stopped due to low sales. Unlike VPD, which serves in humans as a prodrug to VPA (Bialer, 1991), VPD’s constitutional isomer (VCD) acts as a drug on its own, with minimal biotransformation to its corresponding acids and with no evidence of teratogenicity in mice strains that are susceptible to VPA-induced teratogenicity (Bialer & Yagen, 2007).

VCD’s introduction in Europe preceded that of VPA, and published articles from the 1960s attribute VCD’s anxiolytic activity to its chemical similarity to butabarbital or valerian (Roszkowski & Govier, 1962). Figure 2 shows how butabarbital is a chemical origin of VCD and its CNS-active urea derivative valnoctyl urea (Shimshoni et al., 2007). Recently, a successful double-blind controlled clinical trial with racemic VCD in acutely manic patients showed that, in all efficacy measures, VCD (1,200 mg/day; n = 15) was significantly more effective as an add-on to risperidone compared to placebo (n = 17) (Bersudsky et al., 2010). Following this successful study, a phase IIb study with VCD in patients with acute mania will start at the end of 2012.

Therefore, despite the completely different chemical structures of VPA and phenobarbital, the fact that the CNS-active VPA follow-up compound VCD can be formed by a cleavage of the barbiturate butabarbital (Fig. 2) can be regarded as a closing of the circle between the two chemically diverse major AEDs, PB and VPA, introduced independently in 1912 and 1967, respectively.

AEDs Introduced after 1990

The chemical structures of the AEDs introduced after 1990 show a high diversity, and thus they cannot be related to phenobarbital (Bialer, 2006, 2012; Perucca et al., 2007).

As with PB, PHT, CBZ, and VPA, all AEDs introduced after 1990 that are not second-generation drugs (except vigabatrin and tiagabine) were developed empirically (sometimes serendipitously) using mechanism-unbiased anticonvulsant animal models. The empirical (sometimes serendipitous) nature of the discovery of new AEDs in the last three decades, coupled with their multiple (MOAs), explains their diverse chemical structures. It also explains why so far no correlation has been found between the AEDs’ chemical structures and their MOAs.


Target-based drug design or the “targephilia” mantra of: “one gene, one protein, one function” may be useful in developing 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors (statins), HIV protease inhibitors, or antibiotics, but it is not useful for developing antiepileptics or CNS drugs (Enna & Williams, 2009; Bialer & White, 2010). This is because all successful AEDs have multiple MOAs and the two single-mechanism AEDs developed by mechanism-based design are not widely used due to side effects related to their single MOA. In addition, CNS drugs with multiple MOAs have a better probability of being effective in refractory epilepsies and other CNS disorders.

With 16 new AEDs having entered the market in the last 20 years, the antiepileptic market is crowded. Consequently, epilepsy alone is not attractive in 2012 to the pharmaceutical industry for development of new drugs, even though the clinical need of refractory epilepsy remains unmet. Due to this situation, future design of new AEDs must also have potential in nonepileptic CNS disorders such as neuropathic pain, migraine, and bipolar disorder prophylaxis, or fibromyalgia, as demonstrated by the sales revenues of pregabalin, topiramate, and valproic acid. This trend is demonstrated by the recent FDA approval of the GBP prodrug gabapentin enacarbil (Horizant) for treating restless legs syndrome and postherpetic neuralgia. Therefore, the future design of new AEDs must also include a potential in nonepileptic CNS disorders (Mackey, 2010). Two successful phase III clinical trials as add-on therapy in refractory epileptic patients can serve as an entry for regulatory approval and subsequently for gaining a real clinical experience. The successful clinical experience of AEDs in nonepileptic CNS disorders may triple the new CNS drug market potential and will make its costly development worthwhile. This marketing opportunity, coupled with the clinical unmet need patients with refractory epilepsy, provides an incentive for the development new AEDs.

The barbiturates were once widely used as sedative-hypnotic drugs. Except for a few specialized uses, they have been largely replaced by the superior benzodiazepines. In contrast, in epilepsy, phenobarbital is still widely used at present even in developed countries. Yet the development of a nonsedating phenobarbital derivative will answer a clinical unmet need and might make this old class of compounds more attractive.


MB has received in the last 3 years speakers or consultancy fees from BioAvenir, CTS Chemicals, Desitin, Janssen-Cilag, Rekah, Sepracor, Tombotech, UCB Pharma, and Upsher Smith. He has been involved in the design and development of new antiepileptic and CNS drugs as well as new formulations of existing drugs. Parts of this article are included in a previous publication: Bialer M. Adv Drug Deliv Rev 64:887-985 (2012). I confirm that I have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.