In this 5-year period, five major antiepileptic drugs were licensed in Europe and introduced into clinical practice—vigabatrin, oxcarbazepine, lamotrigine, felbamate, and gabapentin (also in this period, Zonisamide was licensed in Japan and South Korea). This was a pace of expansion not experienced before, and it generated a great deal of excitement and promotional activity. Two of the drugs have now largely disappeared from routine practice due to serious toxicity discovered after marketing (felbamate and vigabatrin), and one has fallen out of favor due to a perceived lack of efficacy (gabapentin). Lamotrigine and oxcarbazepine, though, remain in widespread use.
The results of the GABA wave were initially disappointing and rather conflicting. Several potent GABA agonists, for example 4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridin-3-ol (THIP) and muscimol, had proepileptic effects when tested in baboons (Meldrum & Horton, 1978; Pedley et al., 1979). GABA itself could not be used as an antiepileptic as it does not cross the blood–brain barrier, so methods had to be developed to overcome this problem. The first was an attempt to devise what was hoped would be a GABA prodrug, progabide. The second method was more successful. GABA is catabolized at the GABAergic synapse by the enzyme GABA transaminase (GABA-T) and a compound was devised at the Centre de Recherche Merrell International in Strasbourg, which proved to be an irreversible inhibitor of GABA-T. This compound was gamma-vinyl GABA and soon was shown to raise GABA levels in the brain in rodents and in mice (Lippert et al., 1977; Schechter et al., 1977) and then to afford protection in mice against audiogenic seizures. A year later, Meldrum and Horton (1978) reported the effect of this new compound in the photosensitive baboon model. They demonstrated protection at an intravenous dose of 450–950 mg/kg against generalized myoclonus or seizure responses induced by photic stimulation in baboons without or with priming with subconvulsant doses of allylglycine. The protection became maximal 1–3 h after injection, and continued for 7–24 h. Other animal models were studied, and by the early 1980s, 3 years after the discovery of its neurochemical action, vigabatrin (as it was then known) was being used in human studies.
The first reported short-term phase II single-blind efficacy studies were carried out in 1983 by Lennart Gram et al. (1983). About half of 15 patients had a 50% or more reduction in median seizure frequency—a finding that was to be replicated over and over again in more definitive studies. By 1989, four short-term single-blind and six double-blind crossover studies had been carried out in a total of 133 and 141 patients (the crossover studies were reported by Rimmer & Richens, 1984; Gram et al., 1985; Loiseau et al., 1986; Tartara et al., 1986; Remy et al., 1986; Tassinari et al., 1987—a sufficient number and development program in those days for the drug to be licensed in Europe. It was launched in 1989 first in the United Kingdom and then in Denmark, and soon after in 60 countries, although not in the United States. In 1992, Ring and Reynolds (1992) wrote: “The introduction of vigabatrin into clinical practice may prove to be a milestone in the treatment of epilepsy, not only because it is the first novel antiepileptic drug since valproate in the 1970s, but because it appears to be the first successful rational approach to the treatment of epilepsy.” As things turned out, the case of vigabatrin was certainly a milestone, but not in the manner envisaged. Monotherapy trials were quickly carried out and also showed effectiveness (for instance Kälviäinen et al., 1995; Chadwick, 1999).
The rapidity of licensing was perhaps surprising, for there were toxicologic concerns about this drug from the onset. Dose-dependent intramyelinic vacuolation was noticed early on in the initial toxicologic studies of vigabatrin in the 1970s in mice, rats, and dogs, but not in monkeys. And although subsequent studies in humans have not shown any similar effects, the suspicion of neurotoxicity hung over the drug in many quarters right through the 1990s. The mechanism of this unusual effect remains obscure. Over the years, histopathologic examination was carried out in 76 temporal lobe specimens removed at epilepsy surgery, evoked potential examinations in several hundred patients, and serial magnetic resonance imaging (MRI) scans performed in more than 670 patients—without abnormalities being found. Psychosis and other severe psychiatric symptoms were then reported (Sander et al., 1991), and the propensity of the drug to cause marked psychiatric reactions (and aggression) became well established (although, part of this effect was wrongly attributed to forced normalization). The psychiatric effects were hinted at in the double-blind studies but were overlooked. By the mid-1990s, the drug was widely used and was being heavily marketed in Europe—despite unease about neurotoxicity. In the United States, however, the FDA refused initially to license the drug.
Early on in its development, the manufacturers of vigabatrin alighted on the concept that the GABA agonist action was a “rational” design. This became a slogan at the center of a lavish marketing campaign for vigabatrin, the success of which was one reason for the prominent position the drug assumed in the antiepileptic drug market. At that time, few medical meetings were not sponsored by the company, whose hospitality and largesse were overwhelming and whose influence seemed to be everywhere. The claim to be the “first rational antiepileptic” was as illogical then as now, but vigabatrin was certainly “clean” in the sense that its neurochemical effects were seemingly restricted to its binding to GABA-T. By the middle of the 1990s, a rash of papers, supported largely by funding from the manufacturers, appeared showing no human neurotoxicity; an example was the definitive study on CNS conduction times (including visual evoked potentials) of 109 patients followed for 12 months (Mauguière et al., 1997). Then, in January 1997, Mark Lawden and colleagues from Leicester reported severe visual field constriction in three patients taking vigabatrin therapy (Eke et al., 1997). This appeared just as the FDA had decided to issue an approvable letter for the use of vigabatrin the United States. The decision was rapidly retracted; later, the FDA complimented itself on not licensing the drug. A rash of further cases were discovered, and by 1999 it was recognized that vigabatrin causes visual field constriction in around 30–50% of users and that the effects are irreversible. In October 1999, Marion Merrell Dow sent a letter to all prescribers warning of the effect, and sales of the drug began to plummet. In fact, a prescription event monitoring study of vigabatrin conducted between 1991 and 1994 had identified four cases of bilateral, persistent visual field defects for which there was no alternative cause, but this seems to have been neither well publicized nor followed up. The inevitable lawsuits took place to determine whether the company knew or should have known about the risk to vision, and were settled for undisclosed sums.
Once visual field constriction was acknowledged, the licensing authorities heavily restricted the use of vigabatrin, although it remains available for use as adjunctive therapy in partial epilepsy where other options have failed and where the risks of therapy are outweighed by the benefits. In subsequent years, a niche indication in West syndrome, especially where this is caused by tuberous sclerosis, was established (incidentally, one of the few examples where the etiology of an epilepsy syndrome dictates choice of medication). At the time of writing, vigabatrin has become the drug of choice in infantile spasms. And yet in a further twist, in 2006, MRI changes strongly suggestive of intramyelinic edema were observed in the basal ganglia and cerebellum of infants treated with the drug (da Rocha et al., 2006). Further investigation has confirmed this effect in about 11% of all infants (up to the age of 3 years) treated for infantile spasms with high doses of vigabatrin. This was the effect long searched in adults but not found. The changes appear to be reversible, and their significance is currently unclear.
In 1966, Ted Reynolds published a paper postulating that, as some anticonvulsant drugs caused folate deficiency (notably phenytoin and phenobarbitone, an effect noted in the 1950s), perhaps the antifolate effect was also responsible for their antiepileptic effect (Reynolds et al., 1966). Actually, the hypothesis was incorrect, and the antifolate effects of the two drugs have no relevance to the antiepileptic action. Nevertheless, researchers in the Wellcome Research Laboratories in Beckenham, Kent, decided to examine the antiepileptic effect of some antifolate drugs. It was a wild goose chase in terms of mechanism; but the goose laid a golden egg. The phenyltriazine derivatives were among a whole series of compounds studied, and of these, one—BW430C—looked promising, although ironically it too turned out not to have a strong antifolate action. BW430C was then tested in a number of conventional animal models, including the PTZ and MES models. Lamb et al. (1985) within the Wellcome Research Laboratories published comparisons of lamotrigine with the major contemporary drugs: phenytoin, phenobarbital, valproate, carbamazepine, diazepam, troxidone, and ethosuximide, and found lamotrigine to have more effect in abolishing hind-limb extension in the MES than any of these comparators. Inhibition of hind-limb extension is not the same as a human antiepilepsy effect, and indeed, the clinical efficacy of lamotrigine did not seem to match this experimental promise. The drug was then studied in other animal models (thousands of mice, rats, and marmosets participating in the studies) with variable success, and then in human volunteers (Cohen et al., 1985). Toxicology was slow, and consequently the initial clinical studies had to be short-lived. Hemmed in by regulatory restrictions, the company focused on single-dose effects on interictal EEG abnormalities and photosensitivity, and these were interesting and innovative designs (Binnie et al., 1986, 1987). When toxicology was completed, the drug moved into phase III trials—using a crossover design (Binnie et al., 1987; Jawad et al., 1989; Loiseau et al., 1990; Sander et al., 1990; Schapel et al., 1993; Messenheimer et al., 1994). This was an erratic clinical development program, but ultimately successful, and the drug was licensed first in Ireland in 1990, in the United Kingdom in 1991, and then in other countries in Europe. A parallel group study was completed in the United States (216 patients; Matsuo et al., 1993), and the drug was then licensed there as well (the biggest and most lucrative market) in December 1994 as adjunctive treatment for partial seizures with or without secondary generalization in adults.
The crossover designs in this lamotrigine program were used probably for the last time in definitive drug studies, as the FDA began to favor the parallel group formula. Crossover designs have the singular advantage of needing fewer exposed patients to demonstrate efficacy by reducing variance, and remarkably the initial clinical trial program of lamotrigine comprised only 221 patients. In fact, as it turned out, the results of the crossover studies were very similar to those of the parallel group studies, and one wonders whether this cheaper and quicker method of assessment should again be reconsidered.
There have been at least 11 placebo-controlled studies as adjunctive therapy in patients with refractory partial epilepsy, with the early studies using doses guided by blood levels but usually not in excess of 300 mg/day. Several of these early studies showed no significant difference in efficacy between lamotrigine and placebo, but a meta-analysis of all six studies showed a modest effect (an approximately 25% reduction in seizures) in the usual refractory partial-onset epilepsies. This was not a particularly exciting result, but it sufficed to allow licensing. The side-effect profile of lamotrigine was also studied, and the drug proved to be reasonably well-tolerated, with relatively low frequencies of the usual neurotoxic effects. One problem though, not identified in the short-term studies but soon to become very apparent, was the propensity of the drug to cause an allergic rash. This rash occurred in a surprisingly high proportion of initial patients (over 10% in the placebo-controlled trials). Later cases of severe rash were also reported, including Stevens-Johnson reaction, which were occasionally fatal. Several years later the high rash rate was shown to be partly due to the rapid introduction of the drug and could be reduced by slow titration, especially when the drug was used in combination with valproate, which elevated its levels. Nevertheless, serious rash was reported to occur at a rate of about 0.3% even in 1999 (Rzany et al., 1999). The rash was also found to be commoner in children, and this resulted in the regulatory requirement for complex dosing regimens depending on age and co-medication.
Despite what was considered by many to be an initially relatively poor performance, lamotrigine was very heavily marketed as a novel and exciting antiepileptic drug. The marketing was more efficacious than the clinical trials, and there was a rapid growth of sales, especially in Britain and later in Europe. It was approved for use in the United States in 1993, and by 1994 had been used by more than 80,000 people in 40 different countries. Lamotrigine was welcomed as a replacement for the less-marketed older drug leaders—carbamazepine, valproate and phenytoin—at a time when there were few new compounds. Indeed, the introduction of lamotrigine resulted slowly in significant falls in the share of the market occupied by phenytoin (but less so in the United States where fashion dictated support for phenytoin long after it was abandoned in Europe). For over 10 years, from the early 1990s, Wellcome and then in turn its successors GlaxoWellcome and GlaxoSmithKline, dominated medical sponsorship, for instance, in medical meetings (including the ILAE meetings) and in medical journal advertising and sponsored supplements.
There were several other facets of lamotrigine’s development, which at a distance now of 20 years are interesting to reflect on. First, several years after its launch, it became apparent that the drug had value in the treatment not only of the partial epilepsies, its licensed indication, but also in the generalized epilepsies, both primary and secondary. The marketing departments were not slow to exploit this new information, and soon the drug was being hailed as a broad spectrum anticonvulsant. The treatment of generalized epilepsy was dominated by valproate, and an aggressive advertising battle was launched with negative messages made about valproate. In 1993, it was then suggested that valproate caused polycystic ovarian syndrome, a claim based largely on Finnish studies (Isojärvi et al., 1993) and not confirmed in others. Another development was the discovery in the mid-1990s that valproate caused significant teratogenicity. Lamotrigine was then profiled in a new advertising campaign as the drug “for women with epilepsy”—no weight gain, no ovarian cysts, no teratogenicity (in contrast to valproate), and for good measure no interactions with the contraceptive pill (in contrast to carbamazepine). Women with epilepsy became a hot topic, and a remarkable number of publications and conferences were subsequently devoted to the subject. Martha Morrell in New York became a champion and leader in the field. It took another 10 years before the polycystic ovarian story was put into perspective, the remarkable fall in lamotrigine levels in late pregnancy was noted along with interaction with the contraceptive pill, and the drug’s teratogenic potential fully recognized. In fact, at high doses the effects of lamotrigine are probably not dissimilar from those of median doses of valproate. As the patent lapsed, the topic faded from the conference timetables, journals, and bookshops.
In the 1990s, a number of new drugs were licensed, and sales of lamotrigine faltered—not least because of the perception that the drug was a relatively weak antiepileptic. In December 1998, it was licensed for use as monotherapy in the United States, when converting from an enzyme-inducing antiepileptic drug, and in 2004 as monotherapy when converting from valproate. The monotherapy licenses were based on randomized comparative trials in partial-onset seizures, generalized tonic–clonic seizures in adults, the elderly, in children, in Lennox-Gastaut syndrome, and in newly diagnosed patients (Brodie et al., 1995; Reunanen et al., 1996; Besag et al., 1997; Eriksson et al., 1998; Motte et al., 1998; Brodie et al., 1999; Duchowny et al., 1999; Nieto-Barrera et al., 2001;Brodie et al., 2002). An influential double-blind monotherapy comparison with carbamazepine in 260 patients showed lamotrigine to be of equal efficacy but better tolerated (Brodie et al., 1995) and since then lamotrigine has been increasingly used as first-line monotherapy, particularly for generalized epilepsy as an alternative to valproate.
In 2005, as the patent and exclusivity licensing ended, generics moved into the market with a product from TEVA. Since then GlaxoSmithKline sponsorship rapidly diminished, with for instance no satellite symposiums in the recent international and regional congresses. A patent battle between TEVA and Glaxo was initiated in the United States courts, an example of the many internecine legal battles that rage in the pharmaceutical industry.
In 2007, the SANAD study (Marson et al., 2007b), a large-scale pragmatic comparison of initial monotherapy, found lamotrigine to be superior to carbamazepine in partial-onset epilepsy. The results were surprising and contentious, and the pages of the medical journals were filled with critical comment, but it remains to be seen whether this will revive the fortunes of the drug for use in epilepsy at a time when the main thrust of industrial-sponsored research and marketing is directed at its effects in bipolar disease. By December 2007, lamotrigine was licensed in 130 countries and reported to have more than 8.6 million patient-years of experience, and at the time of writing there are 63 current clinical trials in various indications. In terms of unit sales, lamotrigine remains third in the United Kingdom at least after carbamazepine and valproate, with lamotrigine still the most prescribed drug among female teenagers.
The NIH ADD program early on identified one important compound: 2-phenyl-1,3-propanediol dicarbamate (felbamate). It was synthesized in the Wallace Laboratories in the 1950s as a relative of meprobamate in their search for new sedative drugs. It appeared to have little promise as a sedative and so was shelved for many years. Screened by the NIH program, the drug was found to have very low toxicity (“This drug does not kill rats,” as Harvey Kupferberg put it) and had a unique profile of antiepileptic drug activity in both rats and mice (Swinyard et al., 1986). Its pharmacokinetics were rapidly defined, and it soon progressed into clinical trials. In 1991, the first phase II studies, funded by NIH, were reported (Leppik et al., 1991) and were soon followed by two monotherapy studies in 1992–1993. Novel study designs were agreed with the FDA, including studies of felbamate versus no therapy in presurgical patients, and the first regulatory studies in Lennox-Gastaut syndrome. By the mid-1990s, eight major double-blind placebo-controlled studies had been carried out, four with a parallel group add-on design, one a crossover design, and two monotherapy studies. In two studies, an “active control” (low-dose valproate) was employed, with the rationale that it would prevent serious seizure exacerbation yet the superiority of felbamate to be clearly demonstrated. It was a controversial program, but the drug was recommended for approval by the FDA in December 1992. This was the first new antiepileptic drug approved in the United States for 15 years (although in Europe, vigabatrin, clobazam, lamotrigine, oxcarbazepine, and gabapentin, and in Japan zonisamide, had all been licensed), and the move was clearly influenced by general pressure to get an “American” drug to market. The launch of the drug in July 1993 was followed by a massive advertising campaign. The basic message was that here was a new, highly effective American drug, with a remarkable lack of toxicity. Advertisements of a pretty women walking in a flower-strewn meadow with the caption “Seizure control that is easy to live with” appeared in the national press and were highly influential. The culmination, in August 1993, was a Time magazine feature article about a 38-year-old Connecticut homemaker and mother of two boys, named Tiscia, whose uncontrolled seizures and life were dramatically improved by felbamate. Tiscia was quoted “I’m back ! It’s me …. My sister tells me she finally likes me again, She even lent me her car. My mother yells at me again. It’s great.” The campaign was very effective; as Tiscia signed off, the sales of felbamate took off.
In the clinical trials, 1,600 people were exposed to felbamate, over half for 9 months or more, and no significant hematologic or hepatic changes were noted. Within the first year of launch, by August 1994, more than 110,000 patient exposures had resulted. In January 1994, the first cases of aplastic anemia was recorded, and by the end of 1994, 34 cases were reported of which in retrospect 23 were definitely related to felbamate, with 14 deaths. Eighteen patients also developed hepatic failure, with five deaths definitely attributable to felbamate. By August 1994, Wallace laboratories and the FDA recommended suspending use of the drug. The manufacturers then wrote to 240,000 physicians advocating withdrawal unless absolutely necessary. Status epilepticus was precipitated in some cases withdrawing the drug. The planned launch of the drug, about to take place in Europe with a major conference planned in Spain by the European licensee Schering Plough, was cancelled at the last minute. In September 1994 the FDA committee voted (by 6:1) to allow felbamate to remain on the market for restricted cases, with careful monitoring and surveillance and with a black box warning. Since then, the drug has continued to be prescribed, but only in very selected patients. It is currently thought that between 10,000 and 13,000 patients are taking the drug, and since 1995, two new cases of aplastic anemia and two cases of hepatic necrosis have been recorded. The U.S. courts were swamped with more than 100 lawsuits, mostly from people without any bad reaction to the drug but claiming emotional distress or damages for the forced withdrawal. Carter Wallace went out of business, but the drug continued to be manufactured, unpromoted, by MedPoint and now Media Pharmaceuticals. This was an unhappy episode in the recent history of antiepileptic drug therapy. As it turns out, the overall risk of marrow depression is between 27 and 300 per million patients treated (Bialer et al., 2007), and of hepatic failure between one in 26,000–34,000 persons—a risk greater, but not orders of magnitude, so than that of carbamazepine. There are a number of lessons from this debacle. First, the full toxicologic risk of a drug may not be evident on licensing. This is of course obvious, and there are numerous other examples in the antiepileptic drug field where patients take regular therapy for many years and thus where side effects which are rare or which develop only after a long period of exposure or in certain circumstances only (e.g., pregnancy) come to light, sometimes years after licensing. Licensing in this sense must be seen as a balance between allowing undefined risks in order to make available the benefits in terms of seizure control. In the case of felbamate, the short-lived and difficult-to-identify atropalderhyde metabolite of felbamate is probably the causative agent, and susceptible individuals lack the enzyme to metabolize it. It is notable that rodents do not share this metabolic pathway, and so animal toxicology did not detect the risk. Second, it is irresponsible to promote a novel therapy with incautious claims and aggressive advertising—this takes risks with people’s lives. The place of a new antiepileptic may take years to establish, and the excessive marketing of felbamate in its first year resulted in severe restriction and the loss of opportunity to study what might have proved to have been a useful drug in selected circumstances.
Gabapentin is an analog of GABA, the structure of which has been twisted deliberately in the laboratory to make it more lipophilic than the parent compound and thus hopefully able to cross the blood–brain barrier more easily. This was the theory when G. Satzinger, the medicinal chemist working in the company of Goedecke, a subsidiary of Parke-Davis, in Germany first produced the compound. It was developed as a GABA analog, but in fact its antiepileptic effects are due to an entirely different mechanism: to its binding to the α2δ subunit of the neuronal voltage-dependent calcium channel (a fact discovered a decade or more after licensing and launch). Animal and initial patient experience was positive, and in 1984, a clinical development plan was devised in Germany by Bernd Schmidt, who was recruited by Parke-Davis to set up a European clinical research network for the study of the compound. The initial clinical studies were in the field of spasticity and rigidity, as the toxicologic information allowed at this time only 4 weeks of exposure, but later three proof-of-concept (phase IIa) studies were completed in Zurich, Austria/Germany, and in Liverpool, which demonstrated antiepileptic activity. Schmidt recalls that Parke-Davis was at that time uninterested in the drug, partly it was felt at least in Germany because it was “non-NIH” (it was known apparently internally as the Black Forest medication), and partly because they were also handling zonisamide at the time, which took precedence.
The drug “died three times” according to Schmidt, but was resuscitated and then entered into definitive double-blind randomized clinical studies. (UK Gabapentin Study Group 1990; Silvenius et al., 1991; The US Gabapentin Study Group No. 5, 1993;Anhut et al., 1994; Chadwick et al., 1998). These were notable for innovative statistical approaches, and in 1994 gabapentin was licensed in the United States and United Kingdom for use in partial seizures. The drug was also trialed in eight nonepilepsy indications, including major areas such as bipolar disease, mood disorders, and neuralgic pain. over a period of a few years, Neurontin became one of Pfizer’s best-selling products, By 2003 gabapentin was one of the 50 most-prescribed drugs in the United States (its sales rose from $97.5 million in 1995 to nearly $2.7 billion in 2003). According to some estimates, nonepilepsy use accounted for up to 90% of all sales. In 2004, a generic formulation of gabapentin, manufactured by TEVA, was licensed and launched in the United States. In recent years, the drug has been trialed for use in a diverse range of other indications, ranging from the therapy of menopausal hot flushes to the alleviation of drug abuse. In epilepsy, gabapentin has gained a reputation as a rather weak antiepileptic, although well-tolerated, and seems to have a particular role in the elderly, where its gentle nature and lack of drug interactions are particular advantages. It has also been trialed in children (Appleton et al., 1999), but does not seem to have gained much popularity. Its place in the therapy of both pain and epilepsy has to a large extent been superseded by the newer drug pregabalin.
Drugs introduced between 1995 and 2009
In this 15-year period, a further nine new antiepileptic drugs were licensed in either Europe or the United States. It is too early to place these drugs in a historical perspective and they will be briefly listed here.
Topiramate (Topamax), a monosaccharide derived from fructose, was developed initially to be an antidiabetic drug (it has only weak action in this regard) and was then found to have antiepileptic action after routine screening. It has multiple mechanisms of action, including carbonic anhydrase inhibition, and exerts a relatively broad spectrum activity in animal models. It rapidly developed the reputation of being a very powerful antiepileptic drug but which has a high rate of side effects (a reputation which derived I remember initially from a presentation made at the Oslo ILAE International Epilepsy Congress in 1993 by Olaf Henriksen, who showed a slide of a single patient with 13 different side effects). In the celebrated meta-analysis of antiepileptic trials by Marson et al. (1997), topiramate was found to be the drug with the greatest antiepileptic effect. This action has been demonstrated repeatedly in a large variety of controlled studies in various countries, in partial and generalized epilepsy and in adults and children, and in Lennox-Gastaut syndrome (Ben-Menachem et al., 1996; Faught et al., 1996; Sharief et al., 1996; Tassinari et al., 1996; Biton et al., 1999; Elterman et al., 1999; Korean Topiramate Study Group 1999; Sachdeo et al., 1999; Yen et al., 2000). Its poor side-effect profile, though, remains a drawback for the drug. One unusual effect is the non-infrequent development of word-finding difficulties, and other troublesome sequelae include a wide range of other CNS effects, the risk of renal calculi, carbonic anhydrase effects, and weight loss. As a result of findings from dosage studies, the general tendency now is to employ much lower doses than were used in the initial trials (see for instance Gilliam et al., 2003; Wheless et al., 2004), and at the lower doses, side effects are much less troublesome. Nevertheless, the drug is usually reserved, at the time of this writing, as a second-line therapy for resistant cases. Recently topiramate was licensed for migraine, and it may end up being used more often for this condition than for epilepsy.
Tiagabine (Gabitril) is a selective GABA-reuptake blocker that was licensed first in France and then widely in Europe in 1996 and in the United States in 1997, on the basis of positive controlled trial evidence (Richens et al., 1995; Kälviäinen et al., 1996, 1998; Sachdeo et al., 1997;Uthman et al., 1998). It has a pure GABAergic action, but unlike vigabatrin does not affect the retina. It was initially much in favor in northern Europe, but then sales began to founder largely because of the drug’s propensity to cause transient CNS side effects if taken before meals and to induce nonconvulsive seizures or an encephalopathy. It is now little used. At one stage it was suggested that the tiagabine is especially effective in lesional epilepsy, but that seems not to have been borne out in wider practice.
Levetiracetam (Keppra) is, at the time of writing, the most successful of all the newer antiepileptic drugs introduced in this last period of our history. It is one of a large family of pyrrolidone drugs, a drug class pioneered by the chemist Gurgea at UCB (Shorvon, 2001), and has a very close structural similarity to piracetam. Early studies in other indications used the racemic mixture, etiracetam. Levetiracetam is the L-enantiomer of etiracetam (the R-enantiomer being an inactive substance in models of epilepsy) and was known in those early days by the code name ucb-L059. It was first investigated in the early 1980s as a drug with cognitive-enhancing and anxiolytic effects. More than 2,000 patients were included in these early studies, the majority receiving doses ranging from 250–1,000 mg/day, but the findings were disappointing. Attention then switched to the field of epilepsy and studies were initiated, not in the MES and PTZ models, which were the usual screening tests employed for instance by the NIH ADD program, and in which the drug has no positive action, but in the amygdala kindling and photosensitivity models. In these alternative models, the drug had excellent efficacy, and clinical trials as adjunctive therapy in the treatment of partial-onset seizures were begun in 1991. I remember its first trials at the Chalfont Centre for Epilepsy, where as principal investigator I was struck by how novel and effective this drug seemed—and substantially better than others being also trialed. The trials were concluded slowly (Ben-Menachem & Falter, 2000;Cereghino et al., 2000; Shorvon et al., 2000), but were eventually completed and were positive. On the basis of these, levetiracetam was licensed in 1999 in the United States and 2000 in Europe. It was marketed under the trade name Keppra (named after the Egyptian sun god; the launch of the compound took place in Cairo, with a conference full of razzmatazz, dinners in tents in the desert, and camel rides). Its mode of action was initially unclear, but in 2004, it was found to bind selectively and with high affinity to a synaptic vesicle protein known as SV2A, which is involved in synaptic vesicle exocytosis and presynaptic neurotransmitter release. This is a novel binding site (shared only by other pyrrolidone drugs, including piracetam), and exactly how binding confers antiepileptic action is unknown. What is clear though is that this is a new and powerful antiepileptic compound whose clinical effects are distinctive—it has action against many types of generalized as well as partial seizures and controls seizures in many patients in whom other drugs are ineffective. It has a generally alerting rather than sedative action—and it is this latter property that is probably the reason for its popularity. Its main side effect is its tendency to induce irritability and occasionally severe aggression and marked behavioral changes. What role the drug will eventually have in the therapy of epilepsy is unknown. But on its current form, it seems likely to this author at least to become a first-line therapy, challenging valproate and carbamazepine; the discovery of new side-effects of toxicity, though, could easily challenge this position. A range of other pyrrolidone drugs are now being studied for antiepileptic effect, and it is also possible that these will at some stage supersede the place of levetiracetam. For UCB, a relatively small Belgian pharmaceutical company, levetiracetam has proved to be an enormous money-spinner, making over 1 billion Euros in 2008, and propelling the company into the big league.
Zonisamide (Excegran, Zonegran) is currently available worldwide but has had a rather erratic clinical development. It was discovered by Uno and colleagues in 1972 (Shah et al., 1972) and approved for licensing in Japan and South Korea in 1989, manufactured by Dainippon Pharmaceuticals (which in 2005 merged to become Dainippon Sumitomo Pharma) as Excegran. An attempt to license the drug in Europe or the United States failed at that time, on the basis of a clinical trial program that was felt to be inadequate by the licensing authorities and also anxieties about the risk of renal calculi. Following a new series of randomized clinical trials in the United States, it was approved there for licensing in 2000, in 2005 in the United Kingdom, and Germany in 2005, and then in many other countries as adjunctive treatment of partial seizures in adults, under the trade name Zonegran. It had in the meantime proved a popular drug in Japan, where it is said now to occupy about 15% of the Japanese antiepileptic drug market, and where it is licensed as adjunctive and monotherapy in partial seizures and generalized seizures. It shows a strong effect in clinical trials (Schmidt et al., 1993; Faught et al., 2001), although side effects are not infrequent. The mechanism by which the drug exerts its antiepileptic action is not entirely clear, and zonisamide has been shown to be an inhibitor of carbonic anhydrase inhibitor, to block repetitive firing of voltage-gated sodium channels, to reduce T-type calcium channel currents, to bind to GABA receptors, and to increase levels of the glutamate transport protein. Its use seems likely to increase.
Pregabalin (Lyrica) was discovered in 1989 by the medicinal chemist Richard Silverman working at Northwestern University. Like gabapentin, pregabalin binds to the α2δ subunit of the neuronal voltage-dependent calcium channel, reduces calcium influx into the nerve terminals, and decreases glutamate release, but it binds much more tightly than gabapentin. Its effect in epilepsy, anxiety, and pain were studied at the Northwestern University and by Pfizer, which licensed pregabalin for manufacture, and the drug was approved for use in the European Union in 2004 for epilepsy, in the United States for epilepsy, diabetic neuropathy pain, and postherpetic neuralgia pain in June 2005, and then in 2007 for fibromyalgia. Its effects in neuropathic pain have been exceptionally profitable, and within 2 years of its launch in the United States, it had brought in $1.2 billion in sales. Interestingly, both Silverman and Northwestern accrued huge sums from the royalties, and pregabalin is a striking example of a drug discovered in an academic rather than industrial setting. This raised the university from 71st to 11th in the league of U.S. Universities’ industrial earnings and has financed a large academic expansion. Universities often led in drug discovery pre-1940, when royalties were modest, but the massive value of the sales of the drug demonstrates how profitable the pharmaceutical industry had become, and now most drug discovery is conducted in-house by the industry; pregabalin is a striking exception. Pregabalin’s effectiveness in epilepsy across the recommended dose range of 150–600 mg daily was shown in three pivotal double-blind, placebo-controlled randomized trials, with higher doses being more effective (Arroyo et al., 2004; Beydoun et al., 2005; Elger et al., 2005). It has the potential for CNS side effects, and its relative place in epilepsy therapy is at present ill- defined.
Stiripentol (Diacomit) has a long history. It is an aromatic alcohol identified as an antiepileptic in 1978. Its initial trials in adults with partial epilepsy were relatively disappointing, complicated by its potent inhibition of the CYP3A4, CYP1A2, and CYP2C19 isoenzymes of the cytochrome P450 system, resulting in marked effects on the levels of other antiepileptic drugs. The first clinical trials were reported in 1984, and the drug mouldered in the background of epilepsy therapy for many years while studies continued largely in France. Interest has been revived by the discovery that stiripentol has a rather marked and seemingly specific action in severe myoclonic epilepsy in infancy (Dravet’s syndrome). As soon as this was established, Biocodex laboratories submitted the drug for approval under the newly created orphan drug scheme of the European Medicines Agency, and stiripentol was licensed in the European Union in 2007 for use, in conjunction with clobazam and valproate, as adjunctive therapy for refractory generalized tonic–clonic seizures in patients with severe myoclonic epilepsy in infancy. This is the first antiepileptic drug to be licensed for such a narrow indication or for a specific syndrome, and it remains to be seen what its ultimate place in therapy will turn out to be.
Rufinamide (Inovelon) was discovered and initially developed in the early 2000s by Novartis, and then certain rights relating to its epilepsy indications were licensed in 2004 to Eisei. It is a triazole derivative, which probably acts by blocking sodium channels. It was initially trialed in adults with partial epilepsy with rather disappointing results. Then a double-blind, placebo-controlled study in children with Lennox-Gastaut syndrome showed a marked clinical benefit—in particular for drop attacks, which are a major source of disability and morbidity in this severe form of epilepsy—and the drug was approved for use in this syndrome under the orphan drug scheme in the European Union in 2007. It is too early to say how useful rufinamide will turn out to be. But Synosia Therapeutics, an American drug development company, signed an exclusive, worldwide licensing agreement (outside Japan) with Novartis to develop and commercialize rufinamide for the treatment of anxiety disorders and bipolar mood disorders, and this may prove to be the major market for the drug. By early 2008, more than 2,500 patient-years of exposure to the drug had been gained in epilepsy studies, and the good safety profile was one reason for commercial interest in pursuing an anxiety indication.
Lacosamide (Vinpat) was initially developed by Schwartz Pharma (under the names SPM 927, ADD 234037, harkeroside, erlosamide) as a drug for the treatment of partial seizures and status epilepticus. It is a synthetic derivative of the amino-acid D-serine. Schwartz was taken over by UCB Pharma in 2006, and in 2008 the drug was licensed in the European Union for adjunctive therapy in partial seizures following a large multicenter, international study (Ben-Menachem et al., 2007). It acts at the sodium channel, but by a different mechanism from carbamazepine or phenytoin. It is also under investigation for the much bigger indication of diabetic neuropathic pain. It was launched in Europe in September 2008.
Other antiepileptic drugs
A number of other drugs are currently in active development. Some are derivatives of existing drugs, hoping to replace the older drugs with their shorter patent life and also to improve features—examples are brivaracetam (ucb 34714) and seletracetam (ucb 44212)—structurally similar to levetiracetam; eslicarbazepine acetate (BIA 2-093) and JZP-4—structurally similar to carbamazepine; valprocemide and propylsopropyl acetamide—structurally similar to valproate; T-2000—structurally similar to phenobarbital; and fluorofelbamate—structurally similar to felbamate. Others are novel structures—examples are 2-deoxy-D-glucose, ganaxolone, Huperzine A, ICSC 700-008, NAX-5055, NS1209, talampanel, tonabersat, and YKP3089. Many of these act at the glutamate receptor (including talampanel, which is an AMPA antagonist). There are few new drugs currently designed for totally new targets, although ganaxolone is a neurosteroid and Huperzine A (derived from a Chinese herbal remedy and classified by the FDA as a dietary supplement) has a number of known actions on transmitters not normally associated with epilepsy. In many cases, the exact mechanisms of action are not known. This is an active pipeline that holds promise for the future, although currently none of the drugs in development looks so strikingly good that a paradigm shift in epilepsy therapy looks likely. The promise of the molecular age of personalized medicine, pharmacogenetically determined therapy, stem cell therapy, or drugs designed for action at completely novel targets is still far away.
Of these drugs, four (retigabine, carisbamate, eslicarbazepine, and brivaracetam) seem likely to be licensed in the next few years. In addition, as in previous decades, there are other compounds that were investigated in the period under study as antiepileptic drugs, and which entered early human studies, but which have not progressed (examples are dezinamide, nafimidone, ralitoline, milacemide, loreclezole, and losigamone). These are also briefly discussed in subsequent text.
Retigabine: (N-[2-amino-4-(4-fluorobenzylamino)-phenyl] carbamic acid ethyl ester) is a truly novel drug directly activating as it does the voltage-gated potassium channels that conduct the M-type potassium current. No other current antiepileptic drug does this, and it has entered phase III clinical trials and shows promise. It seems to be extremely effective in some patients who are intractable to other compounds, and its wider use is, accordingly, awaited with interest. By the same token, the currently disclosed side-effect profile is depressingly similar to that of other antiepileptic drugs.
Carisbamate, previously known as YKP 509; RWJ-333369, is due to be licensed in Europe for the treatment of refractory partial seizures in 2009. It is a carbamate derivative, developed in South Korea by SK Pharma and licensed in 1998 to Johnson & Johnson. Its proposed advantages include very good tolerability at least up to 300 mg, a broad spectrum of activity in preclinical testing, including the GAERS model, action in human photosensitivity, and little effect on the blood levels of other drugs. Importantly, it is also the first licensed compound shown to prevent epilepsy in an experimental post–status epilepticus model. However, it has relatively limited efficacy in clinical trials, and phenytoin greatly increased its clearance.
Eslicarbazepine acetate is a derivative of carbamazepine that produces structurally different metabolites and avoids the 10,11-epoxide stage, which it is hoped will improve tolerability without lowering efficacy. This bold aspiration seems to be realized in the preliminary clinical trials. Studies are ongoing.
Brivaracetam: The enormous success of UCB Pharma with levetiracetam stimulated a search for further—racetam derivatives. Two were found that show great promise—brivaracetam (UCB 34714) and seletracetam (UCB 44212). The company decided to pursue the former, which binds more strongly to the SV2A protein than levetiracetam and also has sodium-channel blocking action. In animal models it appears more potent than levetiracetam but, because it is extensively metabolized in the liver, it has its own pharmacokinetic interactions. It seems in early clinical studies to be highly effective and well tolerated and is an exciting prospect for the future.
Remacemide is an interesting compound, manufactured by AstraZeneca, which was in clinical development for many years and initially showed promise. It, and its active metabolite (F12495AA, FPL12925, AR-12495AA), are N-methyl-d-aspartate (NMDA) receptor antagonists. The drug also blocks voltage-gated sodium channels. It was developed first as an antiepileptic drug, then in two well-conducted and well-controlled randomized clinical trials in newly diagnosed epilepsy, an unusual but welcome departure from the traditional clinical trial route (Brodie et al., 2002; Whitehead et al., 2002). Unfortunately, a Cochrane Review in 2002 concluded that in the two clinical trials of 514 patients there was only a modest effect on seizures and the drug was more likely to be withdrawn than placebo. Its potential as a neuroprotectant in epilepsy has not been pursued, but it is being studied after stroke in an attempt to minimize cerebral damage, and in Parkinson’s disease and Huntington’s disease.
Dezinamide was the name chosen for AHR-11748, the desmethyl metabolite of fluzinamide, manufactured by A. H. Robins Company and developed by Athena Neurosciences. Experimental findings indicate that it has antiepileptic properties, and its pharmacokinetic properties were assessed in healthy volunteers and in phase II studies. But it was not further trialed clinically. Dezinamide had a promising and unusual profile in the NIH ADD program, but it nevertheless probably acts as a sodium-channel blocker. However, it can cause neurotoxic side effects, including hypomania and rash. Its development ceased in the 1990s, although for financial rather than scientific reasons.
Nafimidone was an imidazole derivative, developed by Syntex Research Laboratories and supported by the Epilepsy Branch of NINDS. Early studies showed striking efficacy, and the experimental and clinical profile of the compound was similar to that of phenytoin and carbamazepine. However, it was said to have a very poor therapeutic/toxicity ratio, and further clinical development was abandoned in the early 1990s.
Ralitoline is a thiazolidinone derivative and manufactured by Warner-Lambert. Experimentally, it resembled phenytoin and carbamazepine and was thought to act on the sodium channel. The major problem identified early in development was its very short half-life, and possibly largely for this reason, and despite a rather promising efficacy profile, its clinical development stopped in the early 1990s.
Milacemide was designed as a glycine agonist and sponsored by Monsanto-Searle. Although open studies showed effectiveness, early double-blind studies found little antiepileptic efficacy and the drug was abandoned as an antiepileptic. Its development continued in other indications, including myoclonus, but was effectively abandoned by the year 2000.
Loreclezole: This triazole derivative, sponsored by Janssen Pharmaceutical, has an experimental profile similar to that of the barbiturates and benzodiazepines, and was found to have promising results, both in monotherapy and as add-on therapy in phase II trials. It binds to the same site as the active ingredient, valerenic acid, of the valerian plant, which is a well-established traditional herbal remedy for epilepsy. Loreclezole has been a valuable, indeed almost cult, compound used to elucidate the functioning of the GABA receptor because of its unique binding properties. Although it continues to be studied experimentally, it seems not to be entering any large-scale clinical development program.
Losigamone was developed by Willmar Schwabe in the late 1980s, in conjunction with the NIH ADD program. It has continued to be investigated in a low-level way, and is now in phase III studies; whether it will ever be licensed after such a long gestation remains to be seen. It has been postulated that it blocks seizure spread by acting on voltage-gated ion channels. Uniquely for an antiepileptic, the drug is a β-methoxy-butenolides and exists as a racemic mixture of two enantiomers (AO-242 and AO-294). In clinical trials, the enantiomer AO-242 seems to be more potent than AO-294 or racemate, and is apparently effective against partial and secondary generalized seizures. A particular advantage is its good tolerability.