Strategies for identifying and developing new anticonvulsant drugs Harvey J. Kupferberg Introduction
- the a m o u n t of compounds available;
The search for new antiepileptic drugs requires the appropriate choice of animal models for the identification of pharmacologic and toxicologic activity as well as new mechanisms of action. A criticism often raised when discussing a new animal model of epilepsy is that they are validated using the current therapeutic agents. Unfortunately, there is no single animal model that reflects all of the pathophysiological processes and symptomatologies of the epilepsies. In fact, there are seizure types and symptomatologies t h a t are not associated with an animal model. The ideal animal model in the search for new therapeutic agents would be one t h a t can identify all drugs active in all forms of epilepsy, independent of mechanisms of action. These deficiencies, however, have not deterred the search for new therapeutic agents. The first drugs were identified using a non-mechanistic approach [1-4]. The animal models used for the search of newer anticonvulsant drugs are varied a n d have been the subject of several r e c e n t reviews [5 6]. Different approaches can be used in the discovery of new therapeutic agents for the treatment of seizure disorders by choosing a model: - specific for a seizure type, e.g. complex partial [6 7], absence [8-10] and status epileticus [11]; - for a mechanism of action of neural excitability [12] or inhibition [13]; - t h a t blocks the final expression of the experimental seizure (mechanism-independent model) [4 14]. In any case, several restrictions and conditions are inherent in the choice of model for identifying a new drug: - the number of compounds to be screened;
- the complexity of the animal model and the technical expertise required for its production and use. The advantage of using the mechanism-independent animal model approach to drug discovery is its ability to identify pharmacologically active compounds quickly. This intern permits further synthesis to maximize activity and duration of action. This approach, however, does not give any indication of therapeutic spectrum or mechanism of action of the experimental compound. Once a specific chemical entity has been identified and its chemical structure compared to existing anticonvulsant drugs, more advanced studies can t h a n be u n d e r t a k e n to evaluate these important parameters. Primary screening
A number of animal models are presently available as a primary screen for the identification of anticonvulsant activity. These models identify antiseizure activity of all compounds currently used for the t r e a t m e n t of a wide variety of seizure types. They include several genetic or reflex seizure models of epilepsy as well as electrically and chemically induced seizures. The genetic models of epilepsy include the photosensitive baboon, Papio papio, [15 16], several strains of audiogenic seizure mice (DBA/2J [17], SJL/J [8], Frings and O'Grady [18]), genetically epilepticprone rats (GEPR-3 and GEPR-9 [19 20]), Mongolian gerbil (Meriones unguiculatus [21-23]), chickens [24], tottering mice (tg/tg strain) [25] and dogs [26]. The audiogenic seizure-susceptible mice exhibit wild running, followed by generalized tonic
Kupferberg HJ. Strategies for identifying and developing new anticonvulsant drugs. Pharm Weekbl [Sci] 1992;14(3A):132-8. Abstract
Keywords
Anticonvulsants Disease models, animal Drug screening Epilepsy
H.J. Kupferberg: Preclinical Pharmacology Section Epilepsy Branch, National Institute of Neurological Diseases and Stroke, National Institute of Health, Bethesda, Maryland, USA.
132
The identification of new anticonvulsant drugs depends on the use of different animal models of epilepsy. The models should be mechanism-independent, able to screen a large number of compounds, at limited cost and technical expertise. Primary screening models include genetic or reflex models of epilepsy and electrically and chemically induced seizures. Once active compounds have been identified, more advanced mechanistic and seizure-specific models are needed to refine the choice of a lead compound. These can be either in vivo or in vitro models. Models known to interact with specific receptors or the production of the putative neurotransmitters of neural excitability or inhibition are valuable in assessing possible mechanisms of action. In vitro models have evolved as important tools in correlating changes in electrical phenomena and therapeutic spectrum. The use of the hippocampa] slice and the cultured neuron permits classification of anticonvulsant activity based on cellular actions of the drug. Interactions by the experimental drugs with specific subcellular fractions of the central nervous system augment information on possible mechanisms of action. The final choice of compounds for development requires synthesizing and comparing all of the pharmacodynamic information with the pharmacokinetic and toxicologic data. In the final analysis, no single animal model of epilepsy known today can assure the development of better drugs for all treatment of the epilepsies. Accepted September 1991.
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
Table 1 Effect of anticonvulsant drugs in three genetic models of epilepsy Drugs
Phenobarbital Primidone Phenytoin Carbamazepine Ethosuximide Trimethadione Valproate Diazepam Clonazepam
DBA/2 mice with Rats with audiogenic audiogenic seizure seizure (clonic-tonic) ED50 (mg/kg intraperitoneally)
ED50 (mg/kg orally)
minor seizures
major seizures
2.7 n.d. 2-14 n.d. 130-150 140 55-300 0.04-0.12 0.005
7.2 n.d. 42 14 -40 195 410 n.d. 0.1
14 n.d. > 40 -40 360 n.d. 180 0.38 0.02 intraperitoneally
8.1 11 22 17 700 n.d. 240 0.7 0.03 intraperitoneally
seizures when exposed to high intensity sound. The intensity of sound usually is in the range of 90-120 db with frequencies of sound between 12 and 16 kHz. Seizure susceptibility in some strains is enhanced if the mice are exposed to high-intensity sound in a priming fashion [18]. In addition, seizure susceptibility changes with age usually reaching a m a x i m u m level between week 2 and 4 of life. Similar seizure and anticonvulsant profiles are also seen in the soundinduced seizure GEPR-9 and the tactile-induced seizure Mongolian gerbil. There also appears to be no critical age-dependency of seizure susceptibility in rats. The anticonvulsant activity of several clinically effective therapeutic agents in suppressing sound-induced seizures in these genetic models of epilepsy is shown in Table 1. The major disadvantage of these models is the cost of maintaining the colony. Chemicals t h a t induce seizures can be used as a primary screen of anticonvulsant activity. They produce both clonic and tonic seizures when administered parenterally. A progression of symptoms is seen, depending on the dose and the route of administration. These symptoms follow a pattern of focal activation of the central nervous system, such as myoclonic jerks and clonic spasms, followed by wild running, culminating in a generalized tonic seizure. The most commonly used chemoconvulsant is pentetrazol (pentenyltetrazol, metrazol, leptazol). It can be used as a primary screen for anticonvulsant activity when doses t h a t produce hindlimb tonic extension are administered [27]. A dose of pentetrazol t h a t produces solely clonic seizures can not be used as a primary screen for new compounds, as phenytoin and carbamazepine have no effect on the blocking of clonic seizures [28]. Fluorothyl, a fluorinated derivative of ethylic ether, produces seizures in both men and animals following its inhalation. Simon et al. demonstrated t h a t 18 drugs used for the treatment of a variety of different seizure types were effective in preventing the fluorethyl-initiated tonic seizures and death in mice [29]. 14(3A) 1992
Gerbils with relax seizure (ED50) (mg/kg orally)
Pharmaceutisch Weekblad Scientific edition
Electrically induced seizures are the most frequently used animal model for the identification of anticonvulsant activity. The MES test or the supramaximal seizure p a t t e r n test is commonly used to identify central nervous system active drugs. The electrical current for this test is 50 ~A (60Hz) for 0.2 s delivered via corneal electrodes to CF1 mice [14 30]. This current is 4 to 5 times as large as the threshold current needed to produce the tonic seizure. These specific parameters are used to identify compounds t h a t prevent seizure spread, therapeutic agents for generalized tonicclonic seizures. Therapeutic agents for absence seizures are undetected. However, when the current is lowered to 12 ~A, compounds which modify both seizure spread and seizure threshold are identified [31]. C h o o s i n g a lead c o m p o u n d Once highly active compounds have been identified using the primary screens, more advanced mechanistic and seizure type models are needed to refine the choice of a lead compound. These can be either in vivo or in vitro models. The neuropharmacologic, metabolic and toxicologic profiles of the experimental compound must also be established and quantified before the drug can be given to humans. Several in vivo tests can be used to elucidate a compound's possible mechanism of action. Chemical convulsants known to interact with specific receptors involved with neural excitability are useful for this purpose. Convulsants such as isoniazid [33], 3-mercaptopropionic acid [33] allylglycine [34], thiosemicarbazide [35], and methionine sulfoximine alter the synthesis of GABA, an inhibitory neurotransmitter. Bicuculline produces seizures by competitively antagonizing the action of GABA [31]. Picrotoxin interacts with the chloride ionophore of the GABA/benzodiazepine channel complex [36]. The convulsant betacarboline, DMCM, competitively binds to the benzodiazepine receptor and therefore is thought to be an inverse benzodiazepine receptor agonist [37]. 133
Table 2
Minimal neurotoxicity and anticonvulsant activity of benzodiazepines in mice* Benzodiazepine
Rotorod Test and ED50 (mg/kg) TD50 (mg/kg) MES sc PTZ sc Bicuculline
Chlordiazepoxide 40 Oxazepam 13 Diazepam 7.3 Nitrazepam 1.5 Clonazepam 0.18 Benzodiazepine I 192 Benzodiazepine II 301 Benzodiazepine III 3.8 Benzodiazepine IV 4.6
29.2 81 19 18 93 95 65 60 10
1.6 0.36 0.17 0.087 0.0090 18 4.1 0.15 0.14
32 17 1.2 4.7 0.0086 >300 12 1.4 > 5.0
sc Picrotoxin
sc Strychnine
13 4.5 1.2 2.9 0.043 >300 >300 1.4 1.5
> 25 > 15 13 > 4.0 > 50 >200 >100 > 4.0 > 5.0
*Data from [38].
The choice of seizure endpoint is critical as these chemical convulsants produce almost identically seizure patterns. Those compounds which block the hindlimb tonic extension induced the maximal electrical stimulation will also inhibit the tonic extension produced by the chemical convulsants. A differential pattern of pharmacologic activity can be attained by using clonic seizures as the endpoint. An example of the differential anticonvulsant pattern using the bicuculline, picrotoxin and strychnine induced clonic seizure endpoint is shown in Table 2.5 Clinically available and 4 experimental 1,4-benzodiazepines were compared for neurotoxicity, and ability to abtund tonic seizures induced by supramaximal electroshock and clonic seizures induced by the above 3 chemical convulsants [38]. Chlordiazepoxide, oxazepam, diazepam, nitrazepam and clonazepam were effective in preventing clonic seizures induced by picrotoxin and bicuculline at doses less than the neurotoxic TD50. The MES ED50, for the most part, was usually larger than the TD50. The experimental benzodiazepines exhibited varied anticonvulsant profiles compared to the clinically available benzodiazepines. Benzodiazepine I was essentially devoid of imparting anticonvulsant activity in either the bicuculline or picrotoxin tests. Benzodiazepine II was devoid of antipicrotoxin activity. Benzodiazepine IV was devoid of antibicuculline activity. Finally, benzodiazepine III had a similar profile to the
clinically available benzodiazepines. All compounds blocked clonic seizures induces by pentetrazol. Finally, there was an excellent correlation between the ED50's of the benzodiazepines, as measured by the pentetrazol test and their ability to displace labelled flunitrazepam from the benzodiazepine receptor. These results give rise to the speculation for the heterogeneity of the benzodiazepine receptor. A further example of the use of chemically induced seizures in making decisions on the development of new anticonvulsant drugs is shown in Table 3. Two experimental substances of the NINDS Antiepileptic Drug Development (ADD) Programme, ADD 77300 and ADD 77305, have similar chemical structures. The anticonvulsant activities of these compounds were compared to phenytoin, ethosuximide and valproate using electrically and chemically induced seizures tests. Both compounds act by preventing seizure spread at doses less than the TD50. ADD 77300 is similar to valproate whereas ADD 77305 is similar to phenytoin. A general observation can be made after screening over 13,000 compounds in the NINDS ADD Programme. If a compound blocks clonic seizures induced by bicuculline or picrotoxin, it always demonstrates pharmacologic activity in either the supramaximal MES seizure pattern test or the subcutaneous pentyL enetetrazol clonic seizure test. Compounds that interact with excitatory amino acid receptors are thought to have great
Table 3
Anticonvulsant activity and neurotoxicity of several compounds against chemically induced seizures Substance
ADD 77300 ADD 77305 Valproate Phenytoin Ethosuximide
TD50 (mg/kg) 97.7 69.0 425 66.5 440
ED50 (mg/kg) sc PTZ
sc Bicuculline
sc Picrotoxin sc Strychnine
37.7 NA 272 NA* 130
31.8 NA 149 NA* 459
59.1 NA 360 NA* 243
107.5 NA 293 NA* >400
*not active. 134
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
potential as anticonvulsants [12 37]. Glutamate and aspartate excite neurons and cause increased electrical activity when applied directly to cortical cells. N-methyl-D-asparate (NMDA) is considered an agonist of one of these receptors and produces seizures when given parenterally. Other excitatory amino receptors subtypes, kainate and quisqualate, have also been identified. Clonic seizures followed by tonic fore- and hindlimb extension are produced when NMDA is administered intracerebroventricularly to mice. This response is highly dependent on the concentration of NMDA, the volume and duration of the injection. Table 4 shows the dose-response relationship of 7 anticonvulsant drugs in blocking both the clonic and forelimb tonic seizures induced by intracerebroventricularly administered NMDA. All of the clinically effective drugs block both behavioural responses. The doses required to block the tonic seizures are always less t h a n those needed to block the clonic component. This difference is generally found for all tests where both clonic and tonic phases are quantified. It must be concluded from this data t h a t the use of the intracerebroventricular NMDA test is a mechanism-independent test and offers no information as to a compound's mechanism of action. In further experiments, not shown here, when approximately three times the amount of NMDA is administered intracerebroventricularly, supramaximal type seizures are produced. Under these conditions, phenytoin was not effective in blocking the clonic seizures. Similar results were also found when quisqualate was administered by the intracerebroventricular route of administration. Seizure type-specific m o d e l s The advanced evaluation of a new chemical entity should include models which are thought to be specific for a seizure type. Loscher and Schmidt suggested t h a t the amygdala kindled rat model of epilepsy might be best suited for identifying drugs for the t r e a t m e n t of therapyresistant partial seizures [40]. The behavioural seizures are not attenuated by pure absence drugs such as ethosuximide. The kindling phenomenon can be produced in a variety of ways, both chemically and electrically [41]. The most common way to produce the kindled state is stimulation of a specific area of the amygdala via imTable 4
ED50 of anticonvulsant against NMDA-induced seizures Drug
Phenytoin Carbamazepine Phenobarbital Valproate Ethosuximide Clonazepam
14(3A) 1992
Time of test (h)
1 0.25 2 0.25 1
ED50 (mg/kg) clonic seizures
forelimb tonic extension
8.6 14.4 2,8 146.2 408.1 0.093
0.6 3.1 3.1 82.8 85.2 0.029
Pharmaceutisch Weekblad Scientific edition
Table 5
Comparison of anticonvulsant activity (ED50) in corneal and amygdala kindled rats Drug
Corneal* AmygdalM
Phenytoin Carbamazepine Phenobarbital Ethosuximide Valproate Clonazepam
> 120 30-60 59 10-30 30 30 >1000 ND 242 190 0.78 0.064
*Corneal kindling conditions: 8 mA, 2 s, 60 Hz stimulation twice daily for 7 days. Seizure endpoint: abolition of forelimb clonus (class 4 seizure according to Racine [43]). Drug administered orally. t Data from [6]. planted electrodes [42]. A fixed current is applied until an afterdischarge is produced at the site of stimulation. Repeated daily stimulation produces behavioural changes t h a t can be classified or graded. In the early stages of kindling, animals develop focal clonic seizures. Seizures progress to where the animal rears with loss of balance, appearing to have a generalized seizure. A major disadvantage of this model in the early phases of drug evaluation is t h a t it i s labourintensive and produces a limited n u m b e r of animals for testing. Electrical kindling of rats via corneal electrodes produces a progression of behavioural seizures identical to t h a t seen with stimulation via the amygdala. This procedure lends itself for the screening of a l a r g e r number of compounds but cannot be considered a general screening procedure. A comparison of the effectiveness of 6 clinically effective drugs in both the amygdala and corneal-kindled rat is shown in Table 5. The lack of efficacy of phenytoin to block the focal seizures or the afterdischarge in this model rat is striking. There are newer compounds t h a t have an anticonvulsant profile identical to that of phenytoin, yet, suppress both the behavioural seizures and afterdischarge in non-sedative or neurotoxic doses. In v i t r o tests
Several in vitro tests are useful in revealing mechanisms of action. Isolated hippocampal slices [44 45] and cultured spinal cord neuron [46 47] are being used today in assessing the mechanisms of action of anticonvulsant drugs. These techniques have the major advantages of recording important electrical activity under a variety of experimental conditions and tissue penetration by the drug. The disadvantages include the lack of interactions with other neural circuits found in the intact animal as well as the viability of the tissue over time. A practical disadvantage of these preparations is the solubility of the experimental drug in the bath's media. Although m a n y clinically effective compounds are insoluble in water, when given orally they are absorbed into the blood stream and carried to the 135
site of action by plasma protein. The hippocampal slice has been used as a screen anticonvulsant activity. Piredda et al. [45] studied the effects of ctonazepam, ethosuximide and phenytoin on extracellular burst recordings from mouse hippocampal slices caused by pentetrazol (200 ~g/ml). Clonazepam completely abolished the bursting activity whereas ethosuximide had a consistent effect when the concentration reached 400 ~g/ml. Oliver et al. using guinea-pig hippocampal slices, demonstrated the suppressive effects of phenytoin, diazepam, phenobarbital and mesuximide on the interictal spiking caused by penicillin [45]. Carbamazepine has a strong inhibitory effect on the spontaneous field burst of the CA I region of rat hippocampal slices bathed in low calcium and high magnesium [48]. Mouse neurons in primary dissociated cell culture provide a useful technique for studying the mechanisms of action of anticonvulsant drugs [49 50]. Phenytoin and carbamazepine limit high-frequency repetitive firing of action potentials whereas phenobarbital, clonazepam and diazepam augmented GABA responses. Ethosuximide failed to affect either the sustained repetitive firing or GABA responses, therefore must act by a different mechanism. Recently, ethosuximide was shown to be a selective antagonist of T-type voltage-dependent calcium channels in thalamic neurons [51]. I n vitro receptor binding evaluations can also be used to identify interactions with putative receptors of the central nervous system. Several receptors are thought to play an important role in seizure supression. The ability of an experimental compound to bind the benzodiazepine or GABA receptor [52 53] or inhibit uptake of a d e n o s i n e [54] and GABA into brain synaptosomes constitute some of the in vitro subcellular tests. Many of the compounds t h a t displace flunitrazepam from the benzodiazepine receptor are also potent inhibitors of adenosine uptake into synaptosomes. On the other hand, a n u m b e r of potent experimental anticonvulsants inhibit adenosine uptake yet were not ligands for the benzodiazepine receptor in nanomolar concentrations. The anticonvulsant profile of these compounds are not similar to the benzodiazepines. In other cases, compounds bind to the benzodiazepine receptor in micromolar concentrations with anticonvulsant profiles similar to phenytoin. This observation may support the concept of a micromolar benzodiazepine receptor [56].
P r o c o n v u l s a n t activity Some experimental anticonvulsant compounds appear to be proconvulsant when administered in high doses. These compounds prevent seizure spread in both mice and rats but are ineffective in blocking seizures induced by chemical convulsants. The timed intravenous pentetrazol test is used to evaluate the proconvulsant potential of new compounds. Pentetrazol (0.185 mg/min) is infused into the tail veins of mice. The time to the appearance of the first focal seizure (first twich) and clonic seizures is measured. Proconvulsant compounds require less pentetrazol 136
(mg/kg) to produce these endpoints. Phenytoin and carbamazepine do not affect the amount of convulsant needed to produce these endpoints even though they are ineffective in inhibiting clonic seizures induced by this chemical stimulant.
Preclinical development The choice of a lead compound from a series of compounds depends upon weighing all of the parameters: the intrinsic anticonvulsant activity, therapeutic index, toxicologic profile, duration of action and metabolic pathways of detoxication. Pharmacokinetic and metabolic considerations are essential for the rational use of antiepileptic drugs. Patients with intractable epilepsy usually receive more t h a n one antiepileptic drug. Examples of enzyme induction, inhibition, and changes in protein binding, to name a few, abound in the literature [57]. It is, therefore, necessary to obtain preliminary information on these interactions during the preclinical development of any new anticonvulsant drugs. Changes in the pharmacodynamic values can occur with repeated administration of an experimental compound. Metabolic and central tolerance are frequently encountered with anticonvulsant drugs. Metabolic clearance of carbamazepine occurs within two weeks following initiation of therapy. Tolerance to the anticonvulsant effect of benzodiazepines occurs in man. Inhibition of cytochrome P-450 by specific chemical structures is well-known. These types of compounds effectively block the metabolism of concurrent therapy and cause toxicity. Nafimidone [58 59] and denzimol [60] are examples of this type of interaction. Metabolic interactions can be addressed by evaluating the changes in cytochrome P-450 and a variety of in vitro substrate-enzyme markers of metabolic pathways following repeated administration. Proof t h a t central nervous system tolerance does not occur is extremely difficult. The lack of the production of tolerance following repeated administration may be due to a series of complex factors e.g. dosing interval, inadequate dose. Conclusion The most effective strategy for developing a really new and novel anticonvulsant can not truly be defined. Current mechanistic theories of neural transmission, excitability, and inhibition have not produced a drug which has successfully impacted on the t r e a t m e n t of intractable seizures. This deficiency is due to the lack of definitive models which mimic the h u m a n condition. Epilepsy is a series of syndromes with heterogeneous neuropathologies t h a t are not entirely understood. The mechanistic approach of drug development does not suffer from the 'old models produce old drugs' syndrome. The nonmechanistic approach depends on serendipitous discovery of the 'different' molecule from the total number screened. If by chance a 'new' molecule is found and shown to be succesfull in treating intractable seizures, new models of epilepsy can evolve. Until then, new animal models for
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
e v a l u a t i n g c o m p o u n d s w i l l be e v a l u a t e d u s i n g 'old d r u g s , '
27
References 1 Swinyard EA, Kupferberg HJ. Antiepileptic drugs: detection, quantification and evaluation. Fed Proc 1985;44:2629-33. 2 Merritt HH, Putnam TJ. A new series of anticonvulsant drugs tested by experimental animals. Arch Neurol Psychiatry 1938;39:1003-15. 3 Huot J, Radouco-Thomas S, Radouco-Thomas C. Qualitative and quantitative evaluation of experimentally induced seizures. In: Mercier J, ed. Anticonvulsant drugs. Oxford: Pergamon Press, 1973:123-85. 4 Swinyard EA, Brown WC, Goodman LS. Comparative assay of antiepileptic drugs in mice and rats. J Pharmacol Exp Ther 1952;106-319-30. 5 Fisher RS. Animal models of the epilepsies. Brain Res Rev 1989;14:245-78. 6 Loscher W, Jackel R, Czuczwar SJ. Is amygdalakindled rats a model for drug-resistant partial epilepsy. Exp Neurol 1986;93:211-26. 7 Engel J, Cahan L. Potential relevance of kindling to human partial epilepsy. In: Wada JA, ed. Kindling. New York: Raven Press, 1986:37-51. 8 Coenen AML, Van Luijtelaar ELJM. The WAG/Rij rat model for absence epilepsy: age and sex factors. Epilepsy Res 1987;1:297-301. 9 Snead OC HI. Gamma-aminobutyrate mode] of generalized absence seizures: further characterization and comparison with other absence models. Epilepsia 1988; 361-8. 10 Vergnes M, Marescaux C, Depaulis A, Micheletti G, Warter JM. Spontaneous spike and wave discharges in thalamus and cortex in a rat model of genetic petit-mal like seizures. Exp Neurol 1987;96:127-36. 11 Walton NY, Treiman DM. Experimental secondarily generalized convulsive status epilepticus induced by homocysteine thiolactone. Epilepsy Res 1988;2:79-86. 12 Croucher MJ, Collins JF, Meldrum BS. Anticonvulsant action of excitatory amino acid antagonists. Science 1982;216:899-901. 13 Loscher W. Comparative assay of anticonvulsant and toxic potencies of sixteen GABA-mimetic drugs. Neuropharmacology 1982;21:803-10. 14 Swinyard EA. Laboratory assay of clinically effective antiepileptic drugs. J Am Pharm Assoc 1949;38:201-4. 15 Meldrum BS. Photosensitive epilepsy in Papiopapio as a model for drug studies. In: Cobb WA, Duijn HV, eds. Contemporary clinical neurophysiology. Amsterdam: Elsevier Scientific Publishing Company, 1978:317-22. 16 Meldrum BS, Horton RW, Toseland PA. A primate model for testing anticonvulsant drugs. Arch Neurol 1975;32:289-94. 17 Chapman AG, Croucher MJ, Meldrum BS. Evaluation of anticonvulsant drugs in DBA/2 mice with soundinduced seizures. Arzneimittelforschung 1984;34: 1261-4. 18 Collins RL. Audiogenic seizures. In: Purpura DP, Penry JK, Tower D, Woodbury DM, Walter Red. Experimental models of epilepsy. A manual for the laboratory worker. New York: Raven Press, 1972:347-72. 19 Dailey JW, Jobe PC. Anticonvulsant drugs and the genetically epilepsy prone rat. Fed Proc 1985;44: 2640-44. 20 Laird HW, Dailey JW, Jobe PC. The genetically epilepsy prone rat. In: Jobe PC, Laird HE ed. Neurotransmitters and epilepsy. London: Humana Press, 1984. 21 Loskota WJ, Lomax P, Rich ST. The gerbil as a model for the study of the epilepsies. Epilepsia 1974;15: 109-19. 22 Loscher W, Frey H-H. Evaluation of anticonvulsant drugs in gerbils with reflex epilepsy. Arzneimittelforschung 1984;34:1484-8. 23 Majkowski J, Kaplan H. Value of Mongolian gerbils in antiepileptic drug evaluation. Epilepsia 1983;24: 609-15. 24 Johnson DD, Davis HL. Anticonvulsants in epileptic fowl. Arzneimittelforschung 1984;34:1735-57. 25 Heller AH, Dichter MA, Sidman RL. Anticonvulsant sensitivity of absence seizures in the tottering mouse. Epilepsia 1983;25:24-34. 28 Loscher W, Schwartz-Porsche D, Frey H-H, Schmidt D.
14(3A) 1992
P h a r m a c e u t i s c h Weekblad Scientific edition
28
29
30
31
32 33
34
35
36
37 38
39
40
41 42 43 44
45
46
47
Evaluation of epileptic dogs as an animal model of human epilepsy. Arzneimittelforschung 1985;35:82-7. Goodman LS, Grewal MS, Brown WC, Swinyard EA. Comparison of maximal seizures evoked by pentylenetetrazo! (metrazol) and electroshock in mice, and their modification by anticonvulsants. J Pharmacol Exp Ther 1953;108:168-76. Gladding GD, Kupferberg HJ, Swinyard EA. Antiepileptic drug development program. In: Frey H-H, Janz D, eds. Antiepileptic drugs. Handbook of experimental Pharmacology. Berlin: Springer-Verlag, 1985: 341-7. Simon P, Larousse C, Boissier JR. Anticonvulsant effects: Criteria of extrapolation from animal to man. Flurothyl-induced seizures as an example. In: Tedeschi DH, Tedeschi RE, ed. Importance of fundamental principals in drug evaluation. New York: Raven Press, 1968:433-47. Toman JEP, Swinyard EA, Goodman LS. Properties of maximal seizures and their alteration by anticonvulsant drugs and other agents. J Neurophysiol 1946;9: 231-40. Piredda SG, Woodhead JH, Swinyard EA. Effects of stimulus intensity on the profile of anticonvulsant activity of phenytoin, ethosuximide, and valproate. J Pharmacol Exp Ther 1985;232:741. Naik SR, Guidotti A, Costa E. Central GABA receptor agonists: .comparison of muscimol and baclofen. Neuropharmacol 1976;15:479-484. Loscher W. 3-Mercaptopropionic acid: convulsant properties, effects on enzymes of the gamma-aminobutyrate system in mouse brain and antagonism by certain anticonvulsant drugs, aminooxyacetic acid and gabaculine. Biochem Pharmacol 1979;28:1397-407. Ashton D, Wauquier A. Effects of some antiepileptic, neuroleptic and gabaminergic drugs on convulsions induced by d,/-allylglycine. Pharmacol Biochem Behav 1979;11:221-6. Meldrum BS. Convulsant drugs, anticonvulsants and GABA-mediated neuronal inhibition. In: KrogsgaardLarsen P, Scheel-Kruger J, Kofod H, ed. GABA-Neurotransmitters. Copenhagen: Munksgaard, 1979:390405. Olsen RW. The GABA postsynaptic membrane receptor-ionophore complex. Site of action of convulsant and anticonvulsant drugs. Mol Cell Biochem 1981;39: 261-79. Petersen EN. DMCM: a potent convulsant benzodiazepine receptor ligand. Eur J Pharmacol 1983;94:117-24. Chweh AK, Miyahara JT, Swinyard EA, Kupferberg HJ. Correlations among minimal neurotoxicity, anticonvulsant activity and displacing potencies in tritium labeled flunitrazepam binding of benzodiazepines. Epilepsia 1984;24:668-77. Czuczwar Sj, Frey H-H, Loscher W. Antagonism of Nmethyl-d,/-aspartic acid induced convulsions by antiepileptic drugs and other agents. Eur J Pharmacol 1985;108:273 80. Loscher W, Schmidt D. Which animal models should be used in the search for new antiepileptic drugs? A proposal based on experimental and clinical considerations. Epilepsy Res 1988;2:145-81. McIntyre DC, Racine RJ. Kindling mechanisms: current progress on a experimental epilepsy model. Prog Neurobio] 1986;27:1-12. Goddard GV. The kindling model of epilepsy. Trends Neurosci 1983;6:275-79. Racine R. Kindling: the first decade. Nenrosurg 1978; 3:234-52. Oliver AP, Hoffer BJ, Wyatt RJ. The hippocampal slice. A system for studying the pharmacology of seizures and for screening anticonvulsant drugs. Epilepsia 1977;18(4):543-8. Piredda S, Yonekawa W, Whittingham TS, Kupferberg HJ. Potassium, pentylenetetrazol, and anticonvulsants in mouse hippocampal slices. Epilepsy 1985;26:167-74. MacDonald RL, McLean MJ. Anticonvulsant drugs: Mechanism of action. In: Delgado-Escueta AV, Ward AA, Porter RJ, eds. Basic mechanisms of the epilepsies, molecular and cellular approaches. New York: Raven Press, 1986;713-6. MacDonald RL, McLean MJ. Mechanisms of anticonvulsant drug action. Electroencephalogr Clin Neurophysiol 1987;39:200-8.
137
48 Hood TW, Siegfried J, Haas HL. Analysis of carbamazepine actions in hippocampal slices of the rat. Cell Mol Neurobiol 1983;3(3):213-22. 49 MacDonald RL. Anticonvulsant and convulsant drug actions on vertebrate neurons in primary dissociated cell culture. In: Schwartzkroin PA, Wheal HV, eds. Electrophysiology of epilepsy. London: Academic Press, 1984:353-7. 50 MacDonald RL, McLean MJ, Skerritt JH. Anticonvulsant drug mechanisms of action. Fed Proc 1985;44: 2634-9. 51 Coulter DA, Huguenard JR, Prince DA. Differential effect of petit mal anticonvulsants and convulsants on thalamic neurones - GABA current blockade. Br J Pharmacol 1990;100(4):807-13. 52 Braestrup C, Squires R. Specific benzodiazepine receptors in rat brain characterized by high-affinity [3H]diazepam binding. Proc Natl Acad Sci USA 1977;74: 3805-9. 53 Enna SJ, Snyder SH. Properties of gamma-aminobutyric acid (GABA) receptor bindings in rat brain synaptic membrane fraction. Mol Pharmacol 1975;100: 81-97.
138
54 Phillis JW, Wu PH, Bender AS. Inhibition of adenosine uptake into rat brain synaptosomes by the benzodiazepines. Gen Pharmacol 1981;12:67-70. 55 Phillis JW, O'Regan MH. The role of adenosine in the central actions of the benzodiazepines. Prog Neuropsychopharmacol. Biol Psychiatry 1988;12:389-404. 56 De Lorenzo RJ. Phenytoin. Mechanisms of action. In: Levy R, Mattson R, Meldrum BS, Penry JK, Dreifuss FE, eds. Antiepileptic drugs. New York: Raven Press, 1989:143-58. 57 Perucca E, Richens A. Antiepileptic drug interactions. In: Frey H-H, Janz D, eds. Antiepileptic drugs. Berlin: Springer-Verlag, 1985:831-55. 58 Buhles WC, Wallach MB, Chaplin MD, Treiman DM. Nafimidone. In: Meldrum BS, Porter RJ, eds. New anticonvulsant drugs. London: Libbey, 1986:203-14. 59 Kapetanovic IM, Kupferberg HJ. Nafimidone, an imidazole anticonvulsant, and its metabolite as potent inhibitors of microsomal metabolism of phenytoin and carbamazepine. Drug Metab Dispos 1984;12:560-4. 60 Testa R, Bertin D. Denzimol. In: Meldrum BS, Porter RJ, eds. New anticonvulsant drugs. London: Libbey, 1986:85-102.
P h a r m a c e u t i s c h Weekblad Scientific edition
14(3A) 1992
- the a m o u n t of compounds available;
The search for new antiepileptic drugs requires the appropriate choice of animal models for the identification of pharmacologic and toxicologic activity as well as new mechanisms of action. A criticism often raised when discussing a new animal model of epilepsy is that they are validated using the current therapeutic agents. Unfortunately, there is no single animal model that reflects all of the pathophysiological processes and symptomatologies of the epilepsies. In fact, there are seizure types and symptomatologies t h a t are not associated with an animal model. The ideal animal model in the search for new therapeutic agents would be one t h a t can identify all drugs active in all forms of epilepsy, independent of mechanisms of action. These deficiencies, however, have not deterred the search for new therapeutic agents. The first drugs were identified using a non-mechanistic approach [1-4]. The animal models used for the search of newer anticonvulsant drugs are varied a n d have been the subject of several r e c e n t reviews [5 6]. Different approaches can be used in the discovery of new therapeutic agents for the treatment of seizure disorders by choosing a model: - specific for a seizure type, e.g. complex partial [6 7], absence [8-10] and status epileticus [11]; - for a mechanism of action of neural excitability [12] or inhibition [13]; - t h a t blocks the final expression of the experimental seizure (mechanism-independent model) [4 14]. In any case, several restrictions and conditions are inherent in the choice of model for identifying a new drug: - the number of compounds to be screened;
- the complexity of the animal model and the technical expertise required for its production and use. The advantage of using the mechanism-independent animal model approach to drug discovery is its ability to identify pharmacologically active compounds quickly. This intern permits further synthesis to maximize activity and duration of action. This approach, however, does not give any indication of therapeutic spectrum or mechanism of action of the experimental compound. Once a specific chemical entity has been identified and its chemical structure compared to existing anticonvulsant drugs, more advanced studies can t h a n be u n d e r t a k e n to evaluate these important parameters. Primary screening
A number of animal models are presently available as a primary screen for the identification of anticonvulsant activity. These models identify antiseizure activity of all compounds currently used for the t r e a t m e n t of a wide variety of seizure types. They include several genetic or reflex seizure models of epilepsy as well as electrically and chemically induced seizures. The genetic models of epilepsy include the photosensitive baboon, Papio papio, [15 16], several strains of audiogenic seizure mice (DBA/2J [17], SJL/J [8], Frings and O'Grady [18]), genetically epilepticprone rats (GEPR-3 and GEPR-9 [19 20]), Mongolian gerbil (Meriones unguiculatus [21-23]), chickens [24], tottering mice (tg/tg strain) [25] and dogs [26]. The audiogenic seizure-susceptible mice exhibit wild running, followed by generalized tonic
Kupferberg HJ. Strategies for identifying and developing new anticonvulsant drugs. Pharm Weekbl [Sci] 1992;14(3A):132-8. Abstract
Keywords
Anticonvulsants Disease models, animal Drug screening Epilepsy
H.J. Kupferberg: Preclinical Pharmacology Section Epilepsy Branch, National Institute of Neurological Diseases and Stroke, National Institute of Health, Bethesda, Maryland, USA.
132
The identification of new anticonvulsant drugs depends on the use of different animal models of epilepsy. The models should be mechanism-independent, able to screen a large number of compounds, at limited cost and technical expertise. Primary screening models include genetic or reflex models of epilepsy and electrically and chemically induced seizures. Once active compounds have been identified, more advanced mechanistic and seizure-specific models are needed to refine the choice of a lead compound. These can be either in vivo or in vitro models. Models known to interact with specific receptors or the production of the putative neurotransmitters of neural excitability or inhibition are valuable in assessing possible mechanisms of action. In vitro models have evolved as important tools in correlating changes in electrical phenomena and therapeutic spectrum. The use of the hippocampa] slice and the cultured neuron permits classification of anticonvulsant activity based on cellular actions of the drug. Interactions by the experimental drugs with specific subcellular fractions of the central nervous system augment information on possible mechanisms of action. The final choice of compounds for development requires synthesizing and comparing all of the pharmacodynamic information with the pharmacokinetic and toxicologic data. In the final analysis, no single animal model of epilepsy known today can assure the development of better drugs for all treatment of the epilepsies. Accepted September 1991.
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
Table 1 Effect of anticonvulsant drugs in three genetic models of epilepsy Drugs
Phenobarbital Primidone Phenytoin Carbamazepine Ethosuximide Trimethadione Valproate Diazepam Clonazepam
DBA/2 mice with Rats with audiogenic audiogenic seizure seizure (clonic-tonic) ED50 (mg/kg intraperitoneally)
ED50 (mg/kg orally)
minor seizures
major seizures
2.7 n.d. 2-14 n.d. 130-150 140 55-300 0.04-0.12 0.005
7.2 n.d. 42 14 -40 195 410 n.d. 0.1
14 n.d. > 40 -40 360 n.d. 180 0.38 0.02 intraperitoneally
8.1 11 22 17 700 n.d. 240 0.7 0.03 intraperitoneally
seizures when exposed to high intensity sound. The intensity of sound usually is in the range of 90-120 db with frequencies of sound between 12 and 16 kHz. Seizure susceptibility in some strains is enhanced if the mice are exposed to high-intensity sound in a priming fashion [18]. In addition, seizure susceptibility changes with age usually reaching a m a x i m u m level between week 2 and 4 of life. Similar seizure and anticonvulsant profiles are also seen in the soundinduced seizure GEPR-9 and the tactile-induced seizure Mongolian gerbil. There also appears to be no critical age-dependency of seizure susceptibility in rats. The anticonvulsant activity of several clinically effective therapeutic agents in suppressing sound-induced seizures in these genetic models of epilepsy is shown in Table 1. The major disadvantage of these models is the cost of maintaining the colony. Chemicals t h a t induce seizures can be used as a primary screen of anticonvulsant activity. They produce both clonic and tonic seizures when administered parenterally. A progression of symptoms is seen, depending on the dose and the route of administration. These symptoms follow a pattern of focal activation of the central nervous system, such as myoclonic jerks and clonic spasms, followed by wild running, culminating in a generalized tonic seizure. The most commonly used chemoconvulsant is pentetrazol (pentenyltetrazol, metrazol, leptazol). It can be used as a primary screen for anticonvulsant activity when doses t h a t produce hindlimb tonic extension are administered [27]. A dose of pentetrazol t h a t produces solely clonic seizures can not be used as a primary screen for new compounds, as phenytoin and carbamazepine have no effect on the blocking of clonic seizures [28]. Fluorothyl, a fluorinated derivative of ethylic ether, produces seizures in both men and animals following its inhalation. Simon et al. demonstrated t h a t 18 drugs used for the treatment of a variety of different seizure types were effective in preventing the fluorethyl-initiated tonic seizures and death in mice [29]. 14(3A) 1992
Gerbils with relax seizure (ED50) (mg/kg orally)
Pharmaceutisch Weekblad Scientific edition
Electrically induced seizures are the most frequently used animal model for the identification of anticonvulsant activity. The MES test or the supramaximal seizure p a t t e r n test is commonly used to identify central nervous system active drugs. The electrical current for this test is 50 ~A (60Hz) for 0.2 s delivered via corneal electrodes to CF1 mice [14 30]. This current is 4 to 5 times as large as the threshold current needed to produce the tonic seizure. These specific parameters are used to identify compounds t h a t prevent seizure spread, therapeutic agents for generalized tonicclonic seizures. Therapeutic agents for absence seizures are undetected. However, when the current is lowered to 12 ~A, compounds which modify both seizure spread and seizure threshold are identified [31]. C h o o s i n g a lead c o m p o u n d Once highly active compounds have been identified using the primary screens, more advanced mechanistic and seizure type models are needed to refine the choice of a lead compound. These can be either in vivo or in vitro models. The neuropharmacologic, metabolic and toxicologic profiles of the experimental compound must also be established and quantified before the drug can be given to humans. Several in vivo tests can be used to elucidate a compound's possible mechanism of action. Chemical convulsants known to interact with specific receptors involved with neural excitability are useful for this purpose. Convulsants such as isoniazid [33], 3-mercaptopropionic acid [33] allylglycine [34], thiosemicarbazide [35], and methionine sulfoximine alter the synthesis of GABA, an inhibitory neurotransmitter. Bicuculline produces seizures by competitively antagonizing the action of GABA [31]. Picrotoxin interacts with the chloride ionophore of the GABA/benzodiazepine channel complex [36]. The convulsant betacarboline, DMCM, competitively binds to the benzodiazepine receptor and therefore is thought to be an inverse benzodiazepine receptor agonist [37]. 133
Table 2
Minimal neurotoxicity and anticonvulsant activity of benzodiazepines in mice* Benzodiazepine
Rotorod Test and ED50 (mg/kg) TD50 (mg/kg) MES sc PTZ sc Bicuculline
Chlordiazepoxide 40 Oxazepam 13 Diazepam 7.3 Nitrazepam 1.5 Clonazepam 0.18 Benzodiazepine I 192 Benzodiazepine II 301 Benzodiazepine III 3.8 Benzodiazepine IV 4.6
29.2 81 19 18 93 95 65 60 10
1.6 0.36 0.17 0.087 0.0090 18 4.1 0.15 0.14
32 17 1.2 4.7 0.0086 >300 12 1.4 > 5.0
sc Picrotoxin
sc Strychnine
13 4.5 1.2 2.9 0.043 >300 >300 1.4 1.5
> 25 > 15 13 > 4.0 > 50 >200 >100 > 4.0 > 5.0
*Data from [38].
The choice of seizure endpoint is critical as these chemical convulsants produce almost identically seizure patterns. Those compounds which block the hindlimb tonic extension induced the maximal electrical stimulation will also inhibit the tonic extension produced by the chemical convulsants. A differential pattern of pharmacologic activity can be attained by using clonic seizures as the endpoint. An example of the differential anticonvulsant pattern using the bicuculline, picrotoxin and strychnine induced clonic seizure endpoint is shown in Table 2.5 Clinically available and 4 experimental 1,4-benzodiazepines were compared for neurotoxicity, and ability to abtund tonic seizures induced by supramaximal electroshock and clonic seizures induced by the above 3 chemical convulsants [38]. Chlordiazepoxide, oxazepam, diazepam, nitrazepam and clonazepam were effective in preventing clonic seizures induced by picrotoxin and bicuculline at doses less than the neurotoxic TD50. The MES ED50, for the most part, was usually larger than the TD50. The experimental benzodiazepines exhibited varied anticonvulsant profiles compared to the clinically available benzodiazepines. Benzodiazepine I was essentially devoid of imparting anticonvulsant activity in either the bicuculline or picrotoxin tests. Benzodiazepine II was devoid of antipicrotoxin activity. Benzodiazepine IV was devoid of antibicuculline activity. Finally, benzodiazepine III had a similar profile to the
clinically available benzodiazepines. All compounds blocked clonic seizures induces by pentetrazol. Finally, there was an excellent correlation between the ED50's of the benzodiazepines, as measured by the pentetrazol test and their ability to displace labelled flunitrazepam from the benzodiazepine receptor. These results give rise to the speculation for the heterogeneity of the benzodiazepine receptor. A further example of the use of chemically induced seizures in making decisions on the development of new anticonvulsant drugs is shown in Table 3. Two experimental substances of the NINDS Antiepileptic Drug Development (ADD) Programme, ADD 77300 and ADD 77305, have similar chemical structures. The anticonvulsant activities of these compounds were compared to phenytoin, ethosuximide and valproate using electrically and chemically induced seizures tests. Both compounds act by preventing seizure spread at doses less than the TD50. ADD 77300 is similar to valproate whereas ADD 77305 is similar to phenytoin. A general observation can be made after screening over 13,000 compounds in the NINDS ADD Programme. If a compound blocks clonic seizures induced by bicuculline or picrotoxin, it always demonstrates pharmacologic activity in either the supramaximal MES seizure pattern test or the subcutaneous pentyL enetetrazol clonic seizure test. Compounds that interact with excitatory amino acid receptors are thought to have great
Table 3
Anticonvulsant activity and neurotoxicity of several compounds against chemically induced seizures Substance
ADD 77300 ADD 77305 Valproate Phenytoin Ethosuximide
TD50 (mg/kg) 97.7 69.0 425 66.5 440
ED50 (mg/kg) sc PTZ
sc Bicuculline
sc Picrotoxin sc Strychnine
37.7 NA 272 NA* 130
31.8 NA 149 NA* 459
59.1 NA 360 NA* 243
107.5 NA 293 NA* >400
*not active. 134
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
potential as anticonvulsants [12 37]. Glutamate and aspartate excite neurons and cause increased electrical activity when applied directly to cortical cells. N-methyl-D-asparate (NMDA) is considered an agonist of one of these receptors and produces seizures when given parenterally. Other excitatory amino receptors subtypes, kainate and quisqualate, have also been identified. Clonic seizures followed by tonic fore- and hindlimb extension are produced when NMDA is administered intracerebroventricularly to mice. This response is highly dependent on the concentration of NMDA, the volume and duration of the injection. Table 4 shows the dose-response relationship of 7 anticonvulsant drugs in blocking both the clonic and forelimb tonic seizures induced by intracerebroventricularly administered NMDA. All of the clinically effective drugs block both behavioural responses. The doses required to block the tonic seizures are always less t h a n those needed to block the clonic component. This difference is generally found for all tests where both clonic and tonic phases are quantified. It must be concluded from this data t h a t the use of the intracerebroventricular NMDA test is a mechanism-independent test and offers no information as to a compound's mechanism of action. In further experiments, not shown here, when approximately three times the amount of NMDA is administered intracerebroventricularly, supramaximal type seizures are produced. Under these conditions, phenytoin was not effective in blocking the clonic seizures. Similar results were also found when quisqualate was administered by the intracerebroventricular route of administration. Seizure type-specific m o d e l s The advanced evaluation of a new chemical entity should include models which are thought to be specific for a seizure type. Loscher and Schmidt suggested t h a t the amygdala kindled rat model of epilepsy might be best suited for identifying drugs for the t r e a t m e n t of therapyresistant partial seizures [40]. The behavioural seizures are not attenuated by pure absence drugs such as ethosuximide. The kindling phenomenon can be produced in a variety of ways, both chemically and electrically [41]. The most common way to produce the kindled state is stimulation of a specific area of the amygdala via imTable 4
ED50 of anticonvulsant against NMDA-induced seizures Drug
Phenytoin Carbamazepine Phenobarbital Valproate Ethosuximide Clonazepam
14(3A) 1992
Time of test (h)
1 0.25 2 0.25 1
ED50 (mg/kg) clonic seizures
forelimb tonic extension
8.6 14.4 2,8 146.2 408.1 0.093
0.6 3.1 3.1 82.8 85.2 0.029
Pharmaceutisch Weekblad Scientific edition
Table 5
Comparison of anticonvulsant activity (ED50) in corneal and amygdala kindled rats Drug
Corneal* AmygdalM
Phenytoin Carbamazepine Phenobarbital Ethosuximide Valproate Clonazepam
> 120 30-60 59 10-30 30 30 >1000 ND 242 190 0.78 0.064
*Corneal kindling conditions: 8 mA, 2 s, 60 Hz stimulation twice daily for 7 days. Seizure endpoint: abolition of forelimb clonus (class 4 seizure according to Racine [43]). Drug administered orally. t Data from [6]. planted electrodes [42]. A fixed current is applied until an afterdischarge is produced at the site of stimulation. Repeated daily stimulation produces behavioural changes t h a t can be classified or graded. In the early stages of kindling, animals develop focal clonic seizures. Seizures progress to where the animal rears with loss of balance, appearing to have a generalized seizure. A major disadvantage of this model in the early phases of drug evaluation is t h a t it i s labourintensive and produces a limited n u m b e r of animals for testing. Electrical kindling of rats via corneal electrodes produces a progression of behavioural seizures identical to t h a t seen with stimulation via the amygdala. This procedure lends itself for the screening of a l a r g e r number of compounds but cannot be considered a general screening procedure. A comparison of the effectiveness of 6 clinically effective drugs in both the amygdala and corneal-kindled rat is shown in Table 5. The lack of efficacy of phenytoin to block the focal seizures or the afterdischarge in this model rat is striking. There are newer compounds t h a t have an anticonvulsant profile identical to that of phenytoin, yet, suppress both the behavioural seizures and afterdischarge in non-sedative or neurotoxic doses. In v i t r o tests
Several in vitro tests are useful in revealing mechanisms of action. Isolated hippocampal slices [44 45] and cultured spinal cord neuron [46 47] are being used today in assessing the mechanisms of action of anticonvulsant drugs. These techniques have the major advantages of recording important electrical activity under a variety of experimental conditions and tissue penetration by the drug. The disadvantages include the lack of interactions with other neural circuits found in the intact animal as well as the viability of the tissue over time. A practical disadvantage of these preparations is the solubility of the experimental drug in the bath's media. Although m a n y clinically effective compounds are insoluble in water, when given orally they are absorbed into the blood stream and carried to the 135
site of action by plasma protein. The hippocampal slice has been used as a screen anticonvulsant activity. Piredda et al. [45] studied the effects of ctonazepam, ethosuximide and phenytoin on extracellular burst recordings from mouse hippocampal slices caused by pentetrazol (200 ~g/ml). Clonazepam completely abolished the bursting activity whereas ethosuximide had a consistent effect when the concentration reached 400 ~g/ml. Oliver et al. using guinea-pig hippocampal slices, demonstrated the suppressive effects of phenytoin, diazepam, phenobarbital and mesuximide on the interictal spiking caused by penicillin [45]. Carbamazepine has a strong inhibitory effect on the spontaneous field burst of the CA I region of rat hippocampal slices bathed in low calcium and high magnesium [48]. Mouse neurons in primary dissociated cell culture provide a useful technique for studying the mechanisms of action of anticonvulsant drugs [49 50]. Phenytoin and carbamazepine limit high-frequency repetitive firing of action potentials whereas phenobarbital, clonazepam and diazepam augmented GABA responses. Ethosuximide failed to affect either the sustained repetitive firing or GABA responses, therefore must act by a different mechanism. Recently, ethosuximide was shown to be a selective antagonist of T-type voltage-dependent calcium channels in thalamic neurons [51]. I n vitro receptor binding evaluations can also be used to identify interactions with putative receptors of the central nervous system. Several receptors are thought to play an important role in seizure supression. The ability of an experimental compound to bind the benzodiazepine or GABA receptor [52 53] or inhibit uptake of a d e n o s i n e [54] and GABA into brain synaptosomes constitute some of the in vitro subcellular tests. Many of the compounds t h a t displace flunitrazepam from the benzodiazepine receptor are also potent inhibitors of adenosine uptake into synaptosomes. On the other hand, a n u m b e r of potent experimental anticonvulsants inhibit adenosine uptake yet were not ligands for the benzodiazepine receptor in nanomolar concentrations. The anticonvulsant profile of these compounds are not similar to the benzodiazepines. In other cases, compounds bind to the benzodiazepine receptor in micromolar concentrations with anticonvulsant profiles similar to phenytoin. This observation may support the concept of a micromolar benzodiazepine receptor [56].
P r o c o n v u l s a n t activity Some experimental anticonvulsant compounds appear to be proconvulsant when administered in high doses. These compounds prevent seizure spread in both mice and rats but are ineffective in blocking seizures induced by chemical convulsants. The timed intravenous pentetrazol test is used to evaluate the proconvulsant potential of new compounds. Pentetrazol (0.185 mg/min) is infused into the tail veins of mice. The time to the appearance of the first focal seizure (first twich) and clonic seizures is measured. Proconvulsant compounds require less pentetrazol 136
(mg/kg) to produce these endpoints. Phenytoin and carbamazepine do not affect the amount of convulsant needed to produce these endpoints even though they are ineffective in inhibiting clonic seizures induced by this chemical stimulant.
Preclinical development The choice of a lead compound from a series of compounds depends upon weighing all of the parameters: the intrinsic anticonvulsant activity, therapeutic index, toxicologic profile, duration of action and metabolic pathways of detoxication. Pharmacokinetic and metabolic considerations are essential for the rational use of antiepileptic drugs. Patients with intractable epilepsy usually receive more t h a n one antiepileptic drug. Examples of enzyme induction, inhibition, and changes in protein binding, to name a few, abound in the literature [57]. It is, therefore, necessary to obtain preliminary information on these interactions during the preclinical development of any new anticonvulsant drugs. Changes in the pharmacodynamic values can occur with repeated administration of an experimental compound. Metabolic and central tolerance are frequently encountered with anticonvulsant drugs. Metabolic clearance of carbamazepine occurs within two weeks following initiation of therapy. Tolerance to the anticonvulsant effect of benzodiazepines occurs in man. Inhibition of cytochrome P-450 by specific chemical structures is well-known. These types of compounds effectively block the metabolism of concurrent therapy and cause toxicity. Nafimidone [58 59] and denzimol [60] are examples of this type of interaction. Metabolic interactions can be addressed by evaluating the changes in cytochrome P-450 and a variety of in vitro substrate-enzyme markers of metabolic pathways following repeated administration. Proof t h a t central nervous system tolerance does not occur is extremely difficult. The lack of the production of tolerance following repeated administration may be due to a series of complex factors e.g. dosing interval, inadequate dose. Conclusion The most effective strategy for developing a really new and novel anticonvulsant can not truly be defined. Current mechanistic theories of neural transmission, excitability, and inhibition have not produced a drug which has successfully impacted on the t r e a t m e n t of intractable seizures. This deficiency is due to the lack of definitive models which mimic the h u m a n condition. Epilepsy is a series of syndromes with heterogeneous neuropathologies t h a t are not entirely understood. The mechanistic approach of drug development does not suffer from the 'old models produce old drugs' syndrome. The nonmechanistic approach depends on serendipitous discovery of the 'different' molecule from the total number screened. If by chance a 'new' molecule is found and shown to be succesfull in treating intractable seizures, new models of epilepsy can evolve. Until then, new animal models for
Pharmaceutisch Weekblad Scientific edition
14(3A) 1992
e v a l u a t i n g c o m p o u n d s w i l l be e v a l u a t e d u s i n g 'old d r u g s , '
27
References 1 Swinyard EA, Kupferberg HJ. Antiepileptic drugs: detection, quantification and evaluation. Fed Proc 1985;44:2629-33. 2 Merritt HH, Putnam TJ. A new series of anticonvulsant drugs tested by experimental animals. Arch Neurol Psychiatry 1938;39:1003-15. 3 Huot J, Radouco-Thomas S, Radouco-Thomas C. Qualitative and quantitative evaluation of experimentally induced seizures. In: Mercier J, ed. Anticonvulsant drugs. Oxford: Pergamon Press, 1973:123-85. 4 Swinyard EA, Brown WC, Goodman LS. Comparative assay of antiepileptic drugs in mice and rats. J Pharmacol Exp Ther 1952;106-319-30. 5 Fisher RS. Animal models of the epilepsies. Brain Res Rev 1989;14:245-78. 6 Loscher W, Jackel R, Czuczwar SJ. Is amygdalakindled rats a model for drug-resistant partial epilepsy. Exp Neurol 1986;93:211-26. 7 Engel J, Cahan L. Potential relevance of kindling to human partial epilepsy. In: Wada JA, ed. Kindling. New York: Raven Press, 1986:37-51. 8 Coenen AML, Van Luijtelaar ELJM. The WAG/Rij rat model for absence epilepsy: age and sex factors. Epilepsy Res 1987;1:297-301. 9 Snead OC HI. Gamma-aminobutyrate mode] of generalized absence seizures: further characterization and comparison with other absence models. Epilepsia 1988; 361-8. 10 Vergnes M, Marescaux C, Depaulis A, Micheletti G, Warter JM. Spontaneous spike and wave discharges in thalamus and cortex in a rat model of genetic petit-mal like seizures. Exp Neurol 1987;96:127-36. 11 Walton NY, Treiman DM. Experimental secondarily generalized convulsive status epilepticus induced by homocysteine thiolactone. Epilepsy Res 1988;2:79-86. 12 Croucher MJ, Collins JF, Meldrum BS. Anticonvulsant action of excitatory amino acid antagonists. Science 1982;216:899-901. 13 Loscher W. Comparative assay of anticonvulsant and toxic potencies of sixteen GABA-mimetic drugs. Neuropharmacology 1982;21:803-10. 14 Swinyard EA. Laboratory assay of clinically effective antiepileptic drugs. J Am Pharm Assoc 1949;38:201-4. 15 Meldrum BS. Photosensitive epilepsy in Papiopapio as a model for drug studies. In: Cobb WA, Duijn HV, eds. Contemporary clinical neurophysiology. Amsterdam: Elsevier Scientific Publishing Company, 1978:317-22. 16 Meldrum BS, Horton RW, Toseland PA. A primate model for testing anticonvulsant drugs. Arch Neurol 1975;32:289-94. 17 Chapman AG, Croucher MJ, Meldrum BS. Evaluation of anticonvulsant drugs in DBA/2 mice with soundinduced seizures. Arzneimittelforschung 1984;34: 1261-4. 18 Collins RL. Audiogenic seizures. In: Purpura DP, Penry JK, Tower D, Woodbury DM, Walter Red. Experimental models of epilepsy. A manual for the laboratory worker. New York: Raven Press, 1972:347-72. 19 Dailey JW, Jobe PC. Anticonvulsant drugs and the genetically epilepsy prone rat. Fed Proc 1985;44: 2640-44. 20 Laird HW, Dailey JW, Jobe PC. The genetically epilepsy prone rat. In: Jobe PC, Laird HE ed. Neurotransmitters and epilepsy. London: Humana Press, 1984. 21 Loskota WJ, Lomax P, Rich ST. The gerbil as a model for the study of the epilepsies. Epilepsia 1974;15: 109-19. 22 Loscher W, Frey H-H. Evaluation of anticonvulsant drugs in gerbils with reflex epilepsy. Arzneimittelforschung 1984;34:1484-8. 23 Majkowski J, Kaplan H. Value of Mongolian gerbils in antiepileptic drug evaluation. Epilepsia 1983;24: 609-15. 24 Johnson DD, Davis HL. Anticonvulsants in epileptic fowl. Arzneimittelforschung 1984;34:1735-57. 25 Heller AH, Dichter MA, Sidman RL. Anticonvulsant sensitivity of absence seizures in the tottering mouse. Epilepsia 1983;25:24-34. 28 Loscher W, Schwartz-Porsche D, Frey H-H, Schmidt D.
14(3A) 1992
P h a r m a c e u t i s c h Weekblad Scientific edition
28
29
30
31
32 33
34
35
36
37 38
39
40
41 42 43 44
45
46
47
Evaluation of epileptic dogs as an animal model of human epilepsy. Arzneimittelforschung 1985;35:82-7. Goodman LS, Grewal MS, Brown WC, Swinyard EA. Comparison of maximal seizures evoked by pentylenetetrazo! (metrazol) and electroshock in mice, and their modification by anticonvulsants. J Pharmacol Exp Ther 1953;108:168-76. Gladding GD, Kupferberg HJ, Swinyard EA. Antiepileptic drug development program. In: Frey H-H, Janz D, eds. Antiepileptic drugs. Handbook of experimental Pharmacology. Berlin: Springer-Verlag, 1985: 341-7. Simon P, Larousse C, Boissier JR. Anticonvulsant effects: Criteria of extrapolation from animal to man. Flurothyl-induced seizures as an example. In: Tedeschi DH, Tedeschi RE, ed. Importance of fundamental principals in drug evaluation. New York: Raven Press, 1968:433-47. Toman JEP, Swinyard EA, Goodman LS. Properties of maximal seizures and their alteration by anticonvulsant drugs and other agents. J Neurophysiol 1946;9: 231-40. Piredda SG, Woodhead JH, Swinyard EA. Effects of stimulus intensity on the profile of anticonvulsant activity of phenytoin, ethosuximide, and valproate. J Pharmacol Exp Ther 1985;232:741. Naik SR, Guidotti A, Costa E. Central GABA receptor agonists: .comparison of muscimol and baclofen. Neuropharmacol 1976;15:479-484. Loscher W. 3-Mercaptopropionic acid: convulsant properties, effects on enzymes of the gamma-aminobutyrate system in mouse brain and antagonism by certain anticonvulsant drugs, aminooxyacetic acid and gabaculine. Biochem Pharmacol 1979;28:1397-407. Ashton D, Wauquier A. Effects of some antiepileptic, neuroleptic and gabaminergic drugs on convulsions induced by d,/-allylglycine. Pharmacol Biochem Behav 1979;11:221-6. Meldrum BS. Convulsant drugs, anticonvulsants and GABA-mediated neuronal inhibition. In: KrogsgaardLarsen P, Scheel-Kruger J, Kofod H, ed. GABA-Neurotransmitters. Copenhagen: Munksgaard, 1979:390405. Olsen RW. The GABA postsynaptic membrane receptor-ionophore complex. Site of action of convulsant and anticonvulsant drugs. Mol Cell Biochem 1981;39: 261-79. Petersen EN. DMCM: a potent convulsant benzodiazepine receptor ligand. Eur J Pharmacol 1983;94:117-24. Chweh AK, Miyahara JT, Swinyard EA, Kupferberg HJ. Correlations among minimal neurotoxicity, anticonvulsant activity and displacing potencies in tritium labeled flunitrazepam binding of benzodiazepines. Epilepsia 1984;24:668-77. Czuczwar Sj, Frey H-H, Loscher W. Antagonism of Nmethyl-d,/-aspartic acid induced convulsions by antiepileptic drugs and other agents. Eur J Pharmacol 1985;108:273 80. Loscher W, Schmidt D. Which animal models should be used in the search for new antiepileptic drugs? A proposal based on experimental and clinical considerations. Epilepsy Res 1988;2:145-81. McIntyre DC, Racine RJ. Kindling mechanisms: current progress on a experimental epilepsy model. Prog Neurobio] 1986;27:1-12. Goddard GV. The kindling model of epilepsy. Trends Neurosci 1983;6:275-79. Racine R. Kindling: the first decade. Nenrosurg 1978; 3:234-52. Oliver AP, Hoffer BJ, Wyatt RJ. The hippocampal slice. A system for studying the pharmacology of seizures and for screening anticonvulsant drugs. Epilepsia 1977;18(4):543-8. Piredda S, Yonekawa W, Whittingham TS, Kupferberg HJ. Potassium, pentylenetetrazol, and anticonvulsants in mouse hippocampal slices. Epilepsy 1985;26:167-74. MacDonald RL, McLean MJ. Anticonvulsant drugs: Mechanism of action. In: Delgado-Escueta AV, Ward AA, Porter RJ, eds. Basic mechanisms of the epilepsies, molecular and cellular approaches. New York: Raven Press, 1986;713-6. MacDonald RL, McLean MJ. Mechanisms of anticonvulsant drug action. Electroencephalogr Clin Neurophysiol 1987;39:200-8.
137
48 Hood TW, Siegfried J, Haas HL. Analysis of carbamazepine actions in hippocampal slices of the rat. Cell Mol Neurobiol 1983;3(3):213-22. 49 MacDonald RL. Anticonvulsant and convulsant drug actions on vertebrate neurons in primary dissociated cell culture. In: Schwartzkroin PA, Wheal HV, eds. Electrophysiology of epilepsy. London: Academic Press, 1984:353-7. 50 MacDonald RL, McLean MJ, Skerritt JH. Anticonvulsant drug mechanisms of action. Fed Proc 1985;44: 2634-9. 51 Coulter DA, Huguenard JR, Prince DA. Differential effect of petit mal anticonvulsants and convulsants on thalamic neurones - GABA current blockade. Br J Pharmacol 1990;100(4):807-13. 52 Braestrup C, Squires R. Specific benzodiazepine receptors in rat brain characterized by high-affinity [3H]diazepam binding. Proc Natl Acad Sci USA 1977;74: 3805-9. 53 Enna SJ, Snyder SH. Properties of gamma-aminobutyric acid (GABA) receptor bindings in rat brain synaptic membrane fraction. Mol Pharmacol 1975;100: 81-97.
138
54 Phillis JW, Wu PH, Bender AS. Inhibition of adenosine uptake into rat brain synaptosomes by the benzodiazepines. Gen Pharmacol 1981;12:67-70. 55 Phillis JW, O'Regan MH. The role of adenosine in the central actions of the benzodiazepines. Prog Neuropsychopharmacol. Biol Psychiatry 1988;12:389-404. 56 De Lorenzo RJ. Phenytoin. Mechanisms of action. In: Levy R, Mattson R, Meldrum BS, Penry JK, Dreifuss FE, eds. Antiepileptic drugs. New York: Raven Press, 1989:143-58. 57 Perucca E, Richens A. Antiepileptic drug interactions. In: Frey H-H, Janz D, eds. Antiepileptic drugs. Berlin: Springer-Verlag, 1985:831-55. 58 Buhles WC, Wallach MB, Chaplin MD, Treiman DM. Nafimidone. In: Meldrum BS, Porter RJ, eds. New anticonvulsant drugs. London: Libbey, 1986:203-14. 59 Kapetanovic IM, Kupferberg HJ. Nafimidone, an imidazole anticonvulsant, and its metabolite as potent inhibitors of microsomal metabolism of phenytoin and carbamazepine. Drug Metab Dispos 1984;12:560-4. 60 Testa R, Bertin D. Denzimol. In: Meldrum BS, Porter RJ, eds. New anticonvulsant drugs. London: Libbey, 1986:85-102.
P h a r m a c e u t i s c h Weekblad Scientific edition
14(3A) 1992
