Clln. Phannarokl~l. II (4); JI8-12I. 1990 OJ I 2. S%J/90/OOQ4-{ll I8j$OS.SOJO CI ADIS Prn5 L,mned All n&hls re-5o!"""cG. ~~
Therapeutic Drug Monitoring of Anticonvulsants
State of the Art
Imtiaz A. Choonara and Anders Rane Insti tute of Child Health, Alder Hey Children's Hospital. Liverpool. England, and Division of Oinkal Phannacology, UniversilY Hospital, Uppsala, S.....eden
Contents
Summary .......................................... . .................. 318 I . Efficacy ................................ .3 19 L I The Lower Limit of the Thel'llpeuhc Ranse ..... 319 L.2 The Upper Limit of the Therapeutic Ranse ..................... 320 2. Toxicity ............................................................. ... 320 3. Indications for Therapeulic DruB MonilorinB of Anticonvulsants ... ...................... 321 4. SJ)«ial Oinical Groups ................. ............................. 321 4. 1 Slatus Epilepticu5 ................................................ . .32 1 4.2 Prq.nancy ................ ................ . ........ 322 4.3 Neonates ..................... .. . 322 4.4 Children ................ ...... 323 5. CompliallCe .......................... . ........ 323 6. Saliva ..................................................................................... . ......... 323 7. Individual Anticon vulsants 324 7. 1 Caroomaupine .................. ... 324 7.2 Clonuepam ............................ ................. . .... 324 7.3 Elh{)$uximide ................. 325 7.4 Phenobarbital ... 325 7.S Phenytoin ... 325 7.6 Valproic Acid .................. 326
There is considcrable interindividual variation in the rela lionship betwttn conlro] of seizures and the serum anticonvulsant concenlra lion. The minimum efTtttive serum conce ntration is dependent on the type and severity of th e epilepsy. and varies from patient to patient. The therapeutic range should be used as a guide to adjust th e dose in ordcr to furthe r improve seizure control or reduce toxicity; the laller is more likel y with higher serum concentrations, but can also be prese nt whcn concenlrations are low. A request for the serum conccntralion of an anticonvulsant should be made only for Bood clinkal reasons, and an interpretation of tha I concentration can only be made if all the relevant clinical detai ls are available.
Therapt'ulle Monl1orlng of AnlleOnl ul!.3nts
319
Indications for [he measurement of serum antleOn, ulsant concentrations Include poor seiZUre' control. tOAICII). suspected gross noncompliance. status epllepllcus and the elapse of 2 to " w~ks after the InlII3110n of themp} . Addillonal drug Iherap). prei'lanc) or illness may alter drug dispoSition in a ",ell controlled pallent and therapeutic drug mOnltormg may. there'fore. help to prevent seiZUre'S secondary 10 these changes. The measure'menl of anllconlulsants 10 sahl'a as opposed to serum may be of benefil In some patients.
It IS almost 30 )ears smce Buchthal and cO",orkcfS studied the relatlonshlp ocI"een serum phen) tOIn concentrallons and control of grand mal epllcps} (Buchthal ct a1. 1960). Since then the conccpt of a therapeutic range has ocen extended and questIOned for dlIYerent anllcoO\ulsants and diagnoses. The abllit} to measurt' plasma concentrations ofanllcOO\ ulsants has ocen shown to help In the control of eplicps) (Lund 197-t North et al. 1980: She,""1n et ai. 1973; Wmg & DulY 1989). It has also been shown that the uncnllcal interpretatIOn of routme $Crum anticonvulsant concentrations docs not Improve patu:nt control (Beardsley el ai. 1983: Froscher et ai. 1981; Woo et al. 1988). There IS no therapeu tic range for any anliconl' ulsan t "'hlch IS applicable to all patients. Even the term 'therapeutic range' itself has been quesllonl'd. and SCI eral authors hale suggested 'optlmal range' as an altematlle (Bartels 1980; Re) nolds et al. 1981: Vajda & Alcardl J 983). There arc 2 maIO conSlderauons ",hl'n a palll'nt IS trcatl'd:
oUlpatil'nts "'lIh grand mal epilepsy. control was beSl achleled wllh serum phen}tom concentratIons abole 15 mg,lL. In 86 pallents (no clinical details 8Jvcn). serum phenytOin concentrations werl' below 14 mg/L and none complained of Side dfects. In contrast. 34 patients had concentrations greater than 30 mgJL: of these 17 (50%) had scvere. and 8 (24%) mild. Side effects. The concept of the therapeutiC range (10 to 30 mgJL for phenytom) had been established. Subsequent studies reponed phen) tOIn toXtCII) at concentrations ab()\e 20 mg,l L (Kull et al. 1964) and therefore the therapcuttc range was revised to 10 to 20 msiL. Therapeutic ranges have Slncc been reported for most anticonvulsants. The alll1 of thiS review IS to dISCUSS some of the problems of the concept of a therapeutIc range.
J. Efficacy: can the patient's disease or
Imllal studies WIth phenytOin Invoh'ed pattents wllh $Clerc grand mal epilepsy. Many pattents can oc adequa te!) con trolled with 'subtherapeullc' plasma concentrations. Feldman and Pippenger (1976) reponed 25 pallents who had been seizurefree for at least 2 yea~: 8 of II patienlS treated wllh phen)toln alone had serum concentrations below 10 msiL (range 0.9 to 6.4 mSlL); 7 of 14 pattents treated "11h phen)toln and phenobarbital In combination had scrum concentratIOns of both anticonl'ulsants oclo" 10 mg/L (range 0 to 9.0 mgj L). Scleral other autho~ have reponed that In some pattents control can be achIeved With ·subthera· peUIiC' concentratIOns of pheny tOin (Dawson & JamIeson 1971 ; Turnbull et al. 1984): Ihis IS con·
s~mptoms
oc controlled? 2. TOXICity: can Side effects of the treatment oc mlnlmlscd or prevented? The delelopment of an anal}lIcal assa} for the measurement ofphen}toin m serum led 10 the subsequent clinical studIes of the relationship between the efficacy and toxicit} ofphen}toln and Its concentratIOn m serum. The mitial prospective slud) of J 2 hospllaltsed patients With severe grand mal eplicps} rcrelvmg phenytoin and phenobarbllal showed conlrol on ly when thl' serum phenytom concentratIOn was greater than 10 mg/L (Buchthal et al. 1960). In the same paper II ",as noted that.
In
I. Eff icacy I.l The Lo"er Llm1l of the Therapeutic Range
elm. pnarmacoklllet. 18 (4) 1990
320
5i5ten! with the clinical impression of many practitioners treating palients with epilepsy. One paper describing the use of ethosuximide gave a suggested therapeutic range of40 to 100 mg,! L. bUI reported I patient with a 93% reduction in seizure frequency with a plasma ethosuximide con~nt ralion
of 16.6 mg/L (Browne et al. 1975). A
prospective study of previously untreated patients with epilepsy showed thaI, in many patients. complete control of seizures could be achieved with 'subtherapeul ic' serum concentrations of phenytoin (Callaghan el al. 1985). A recent randornised. controlled trial showed that there is no advantage in increasing the dose of phenytoin or phenobarbital in patients with 'subtherapeutic' serum concentrations who have been seizure free for 3 months or more (Woo et al. 1988). It is clear. therefore. that no universal lower limit exists for the therapeutic range ofanticon vu lsants. Due consideration must be given to the patient's history offits before the drug dose is increased in patients with apparently subtherapeutic drug concentrations.
1.2 The Upper Limit of the Therapeutic Range Although the initial repons involved patients with severe grand mal epilepsy, subsequent studies have shown that some patients require serum concentrations above the upper limit of the therapeutic range. One long term prospt(:tive study showed considerable interindividual variat ion in the therapeutic serum phenytoin concentration for each patient (lund 1974): several patients with severe grand mal epilepsy required serum concentrations above 20 mg/l to achieve satisfactory control. Although most prospective studies of efficacy and serum concentrations involved patients with grand mal seizures, some included individuals with partial seizures. In some studies. no correlation was found between the serum concen trations ofphenytoin (Gannaway & Mawer 1981: Turnbull et al. 1984), carbamazepine (Strandjord & Johannessen 1980) or valproic acid (Turnbull et at. 1983) and the control of partial seizures. A prospt(:tive study of 40 patients with grand mal epilepsy and 38 with
partial seizures with or without grand mal epilepsy showed that higher plasma concentrations of phenytoin, phenobarbital and carbamazepine were needed for complete control in those with partial seizures (Schmidt et at. 1986). The upper limit should. therefore. be considered as a guide. If clinically indicated it may be necessary to exceed it. It is important to be aware. however. that seizures can be induced by anticonvulsants: this phenomenon has been reported following therapy with phenytoin (Levy & Fenichel 1965). mephenytoin (Troupin & Ojemann 1975), carbamazepine (Shields & Saslow 1983: Snead & Hosey 1985) and diazepam (Prior et a!. 1972). With phenytoin and mephenytoin, these cases are usually (but not always) associated with a high serum drug concentration (Stilman & Masdeu 198 5: Troupin & Ojemann 1975). Therefore, in a poorly controlled patient with a high serum phenytoin concentration. the elinician should consider both an increase and a decrease in the dose as therapeutic possibilities.
1. Toxicity The original studies with phenytoin (Buchthal et al. 1960: Kutl et al. 1964) suggesled a clear relationship between serum concentration and toxicity. Buchthal and co-workers reported no side effccts in patients with serum phenytoin concentrations below 14 mg/l. and severe toxicity was present only with concentrations greater than 30 mg/L. Kutt and co-workers reported nystagmus with seru m phenytoin concentrations of 15 10 25 mg/l. ataxia with concentrations of 30 to 45 mg/ L and mental slowing (confusion, disorientation and lethargy) with concentrations above 40 mg/L. Subsequent studies supported these findings(Haerer & Grace J 969) and free phenytoin concentrations were shown to have an even closer correlation with toxici ty (Booker & Darcey 1973). Side effects of anticonvulsants aTe common (Choonara 1988a) and the reported incidence depends on the thoroughness of the $Carch for them (Herranz et al. 1982). The original studies of phenytoin toxicity (Buchthal et al. 1960) did not
TherapeullC Momlonng or Antlcon\ ulsanl~
tf) to dr/eet side effects. bu t relied on those that the pallent complamed or. Moreover. gi ngival hypenroph) was specificall) e"dudcd. T he more subtlt' eflects ofantlconvulsants on cogni tive function (Thompson & Trimblc (982) were nOt realised at the IImc. SIde effects arc dose dependent. allergic or Id io. s)ncrnllc. \\ hcrcas one "ould C,\pect a correlatIOn bet"ecn to'ICII) and serum concentrallon for thc first menlloned. the last 2 are unrelated to serum concentration. It IS therefore not surpnSlng that somc groups ha\e reponed a correlation octwttn serum concentration and some adverse drug react ions. e.g. ski n reactions (ChadwIck et al. 1984; Wilson et al. 1978) and cardiotoxlci ty (Durelll ct al. (985). "hcreas others have found none (Choonara 1988b; Gram et al. 1979; Herranz et al. 1982). Side effects ma~ also be related to the rormatlon ofa mNaboh te (Park & Kittenngham 1987) "hich has been proposed for carbamazepme-IO.II-epo,Ide (Patsalos et a1. 1985: Schoeman et al. 19!{4). J ust as the ther::lpcutlc rangc must be considered as a general gUidc fo r cflicac~. so II must be realised that to"lelt) can occur with subtherapeutlc or therapeutic concentrallons. and that some patients can tolerate coneenualions above the therapeullc range without an) SIde effects.
3. In dications for Th erapeutic Drug Monitoring of Anticonl'uisants The points abo\e arc mentioned as an argumenl for the appropnate usc of therapeutIc monltormg faclh\les, and to emphasise thc problems of a dogmatIc attitude towards serum concentration anal)scs. There must be a chnlcal rcason to rcqucst a scrum concentration of an antlcon\UlsanL routine monltonng of anllcon, ulsants on all outpallents has no ,alue. eSpt'c13l1) If the patient's sclzures arc "ell controlled. The follOWing arc the most acceptable reasons for requesting the scrum concentratIOn anal)sls of an antlcon, ulsant: I. Poor sclzure Control. 2. TOXlcll) . 3. Suspected gross noncompliance.
321
4. Status epilepllcus. 5. At 2 to 4weeks after Initiation o f therap y. 6. Modification of concomitant drug therapy (additions. dose changes. discon tinuation). 7. Pregnancy. 8. Itlness (fever. hepatIC or renal malfunction). 9. Research. ThiS an lde does not deal wllh drug In teractions between ant lconvulsants as the au thors are unaware of a n) cll mcal Indication fo r treatment with more than I anticonvulsan t at a time; on the contrary. there h ave been a num ber of re ports which have shown that I drug alone is superio r t o comblOallon therap} In terms of effieac) (Lund 1974: Reynolds & Shorvon 1974). Most studies have tn· vol'OO collectIOn of trough samples. i.e. prior to the adm lOlstra tlon of a dose: this ISpreferable. bu t often ImpractIcal tn clinical Sllua tlons. It IS there· fo re essenual to know the lime of admtn tstrat lon of the last dose and thiS should be noted on the analYSIS request form . It takes a period equI\alent to 5 half-Ines before steady-state concentratIons arc reached and tn the case of carbamazeptne. beea usc of ItS au tolnductlon of Its own meta bol ism. Itta kes even longer. The lime reiallonshi p between an cffeclne serum concentration and efficacy has not been established e'cept tn cases of status epilepticus. and the authors therefore "ould recommend a minImum of7to 14 days' admtnistration of a ny anticonvulsan t before the serum eoncentrallon is checked (unless the case IS one of sta tus eptleptlcus). In the casc of phenobarbital it rna)' take up to 4 weeks for steady-state levels to be achie"ed because of the long half-life of that drug.
4. Special Clinical Groups 4.1 Status Epilepllcus Intra,enous d13zepam or clonazepam are the drugs of chOIce for acute SClZures. They ar(: gl\en as a bolus and rna) be conllnued as an InfUSIOn. .,,),5 both cfficac) (termlnallon of the SClzu res) and tOXIClt} (respirator) depression) are Immediate. there IS no need \0 measure theIr serum coneentrallons In the short term. Ifa benzodlazeptne fa ds.
Om Phurmuro/..m l'l 18 (4) /990
322
ph~nytoin is often used: a loading dose of ph~ny toin 15 mg/kg administered as an Infusion la, a ra te of 40 to 50 mg/min in adults or I to 3 mg/ kg/min in children (Albani 1977») in patients not previously rea:iving the drug should result in plasma phenytoin conct'ntrations above 10 mg/L within 10 minutes (Salem et al. 1981). The response to phenytoin in status epilepticus appears to b(: related to the underlying disease. and hen~ widely varying response may b(: seen in different patient groups. Two separate studies have shown a similar serum phenytoin concen tration in responders and nonresponders (Cranford et a\. 1979; Wallis et a!. 1968). Because of the high morbidity and mortality associated with status epilepticus. it is worthwhile to check the serum phenytoin concentration within 30 minutes of the end of the infusion to ensure the concentration is at least 15 to 20 mg/L (Wallis et a!. 1968). If the patient continues to have seizures despite such a serum concentration. it is probably advisable to try another anticonvulsant rather than to con tinue with further doses of phenytoin . Paraldehyde and valproic acid are also effecti ve in status epilepticus. but there is insufficien t data to indicate the need for measurement of serum con~ntrations of these drugs.
4.2 Pregnancy Many women with epilepsy experience an increase in seizure frequency during pregnancy (Knight & Rhind 1975; Schmidt et a1. 1983). Numerous physiological changes occur during pregnancy: the metabolic clearance of drugs appears to be enhanced (Dam et al. 1979; HOgstedt et al. 1983), and there is also an increase in the unbound fraclion of drug (Chen et al. 1982). Despite this increase. the free serum concentrallon of phenytoin IS decreased in late pregnancy (Chen et al. 1982). It has been suggested that monitoring of saliva phenytoin concentrations be carried ou t during pregnancy and in the postnatal period (KnOll et al. 1986). as these correlate with the free plasma concentration. Further prospective studies are required to confirm this suggcstion. In the case of
carbamazeplne there is a small increase In the unbound fracllon and ils metabolite carbamazeplne. 10, II -epoxide. and an increase in the ralio ofepox. ide to carbamazepine (Yerby et al. 1985). suggesting a higher bioactivation rate in prqnancy. Even though pregnancy may alter seizure thresholds as well as drug disposition. and the ther· apeutic range during pregnancy has not 1>«n d~ tennined. until such studies han 1>«n perfonned the aim should be to keep anticonvulsant concentrations at the upper end of the therapeutic range or al the level that has controlled each individual patient prior to the pregnancy. There is greater risk associated with insufficient seizure con trol than of a malformation of the child occurring as a result of treatment. 4.3 Neonates Phenobarbital has 1>«n the drug ofchol~ In the neonatal period. &vcral studies suggest thai control can be ach ieved wi th serum concentrations of 15to 30 mg/l (Pippenger & Rosen 197.5). 12 to]O mg/l (Jailing 19 75) and 17 to 25 mg/l (Lockman et al. 1979). Gal and co-workers used a loading dose of 15 mg/kg followed by additional doses of 5 to 10 mg/kg every 5 to 10 minutes until control was achieved (maximum dose 40 mg/kg) in a group of 71 neonates (Gal el al. 1982). They controlled seizures in 60 (85%) of the neonates. of whom ]6 (60%) responded to plasma conctn trations below 20 mg/ land 17 (28%) 10 concentrations of 20 to 30 mg/ L. Seven neonates ( 12%) responded on ly to plasma concen trations above 30 mg/L. A recent study shows that there is little value in funher doses of phenobarbital once a concentration of 40 mg/l is achieved (Gilman et al. 1989). Sedation. poor fceding and hypotonia have be(:n reported at conctntrations above 50 mg/l (Gilman et al. 1989).60 mg/l (Pippenger & Rosen 1975) or 70 mg/l (Painter et al. 1978). Studies of the pharmacokinetics of phenytoin and valproic acid in the neonatal period havc also been carried out. but there is insufficient data to give a therapeutic range in this agc group.
Therapeutic Monllonng of AnliconvulsaniS
4.4 Children Many of the initial studies Involved children as well as adult patients. and several groups have looked specifically at children. Borofsky and coworkers (1972) reported control in children of all ages with serum phen)toin concentra tions of I to 18 mglL Drug to'\ICIt) (gross ataxia) was reponed in 8 of the 53 childrcn with serum phenytoin concenlraBons ranging from 17.8 to 48 mg/L The same researchers late; subdivided these 53 children into 10 who showed a d,rcct correlallon between seIZure control and serum concentrallon. 8 who failed to achle\e control despJle a high serum concenlration and 30 who were controlled on thc Imtlal dose of phenytOin (Borofsky et al. 1973). Dawson and Jamieson (1971) showed that monitonng serum phenytoin concentralions and subsequently in_ creasing the dose led to improved seizure control in II of 30 children. In a study of 109 children (Norell et al. 1975), serum phenytoin concentrations in those who had been seizure free for a minimum of 3 months ranged from I to 28 mglL In 9 children seizures were only controlled when serum phenytoin concentrations wcrc above 12 mgl L These studies suggest that there is as much interindividual variation in Ihc therapeutic serum concentration ofphcll) tOIn In children as In adults. The availabllll) of modern telemetry facII,tlcs has permlttcd the obJcctlve assessment of EEG·venfied seizures In relation to the serum antIconvulsant conccntratlOn (Braathen et al. 1988: St;\hl et al. 1983). There have been many studies of the use of phenobarbital. valproic acid and other anticonvulsan ls in chi ldren with febri le seizures. although it has been suggested that there is little justification for using these potentially toxic drugs for a condition with such a good long term prognosis when rectal diazepam is available [see the rCVlew by Knudsen (1988)].
5. Complianu Although the measurement of serum an IIconvulsant concentrations can assist In detecting gross noncompliance (concentrations below thc limll of
323
detection suggest that the patient is not laking the medication), other factors may also explain low serum concentrations: examples are low dose. increased metabolism or malabsorption. Carefully worded questions. for example ' How often do you forget to take rour medicineT. 'At what time of the day are you most likely to forget?'. which do not imply criticism. should be used 10 detecting noncompliance (Mucklow & DolleT)' 1978: Pearson 1982). Such quesllons are more uscful than routine monitoring of serum anticonvulsant concentrations In all outpatients. In a prospective study of 96 children. 18 gave a history of noncompliance whereas rou!Jne measurement of serum concentratIOns dctected gross noncompli. ance III on I) 3 (2 of whom gave a positive history of noncompliance). and r::llsed the POSSibI lity In 4 (3 of whom sa"e a positive history of noncompli_ ance) [Choonara 1988b]. Measurement of serum anticonvulsant concentrations to check for noncompliance should only be considered in patients with poor control who den) it despite careful questioning.
6. SaJil'a Saliva is a low protein fluid. and saliva drug concentrations therefore reflect Ihe amount of un· bound drug. This r('lallonshlp has been documented for carbamazepine. phenyt01l1. phenobar. b,tal and pnmldone (McAultfTc Ct al. 1977). The protein b1l1ding of ethosuximldc is insignificant. and therefore salivary and serum concentrations are considered to be the same. The pKa of phen. obarbital is similar to the pH of saliva. and it has been suggested that pH be measured simultan· eously (McAul1ife ct a1. 1977). but this has been disputed b) others (Fnedman et al. 1981: Tokugawa et al. 1986). Valproie acid has a low distribution into sall\ a. and thus measurement of salivaT) concentrations IS of no value (Fung & Ueda 1982). However. the collcction of saliva does not In\ohe \enepunc(urc. and thiS may be of value in some children. Most studIes have correlated salivary concentratIOns wllh serum concentrations without regard
Clm Pharmaroklnf'f 18 (4)
"""
,-, ""'"
CarblllNlz. . . .
~"
.
1. 13
23
1).720
"
CIonaz~m
EthOSlI~lmide
Pt\enobaJb;t.,
-"~
,..,.
'&'101 3·22 ' , '5 '.18
I b c
No'
"
3S
" "
""
R.ler--=StrandfOtd &. JohannesMII (1980) Cahghan.l .,. (1978) Dr~tuss .1 ., (1975) N.i1o" aI . ('987) BrOWN.' II. (1915) 8ucnltlal &. SVlln,m.rtI ('971) FeICIlNln &. Pippenglf
(1976)
Sorolsky lit .,.
(1972)
Number 0 1 controlled patient, In study
,gIL_ Not II receiylI'Ig monolheriPY
for efficacy or 10)!:iciIY. Rylance and Moreland ~tween saliva carbamazepine and phen ytoin conCtntrations and toxicity, but no correlation with efficacy. They sug· gested a therapeutic range for carbamazepine in sa· liva of 1.2 to 3.5 mg/L and for phenytoin ~twttn 5 and 25 mg/L. Nevenheless. the value of moni· toring drug conCtntralion in saliva in the clinical situalion has not been fully assessed, and fun her studies are required.
(1981) found a posilive correlation
7. Individual .4nliconvulsants Table I shows therapeutic ranges for patien ts reo ceiving monotherapy with the drugs reviewed ~Iow.
7.1 Carbamazepine Carbamazepine is the drug of choice for partial com pie)!: seizures. grand mal and temporal lo~ epilepsy. It is as effective as phenytoin butlcss toxic (Callaghan et al. 1985; Cereghino et al. 1974; Shor· von el al. 1978: Troupin et al. 1977). Initial studies suggested a therapeutic range above 4 mg/l (Dam
f99{)
et al. 1975) and between J and 12 mg/L (Strand· jord &. Johannessen 1980). There is. howe\·er. con· siderable interindividual vanatlon in the serum concentration of carbamaztpme reqUired to achie\'e control. In 2 separate prospectIVe studies. complete control was achieved in 5 of25 pat,ents with serum concentrations below 4 mg/l (Shorvon et al. 1978) and in 13 of 28 patients with concentrations of I to 8 mg/l (Callaghan et al. 1978). Childrcn with low serum concentrations of carba mazepine who are nevertheless well controlled have been reponed by Rane et al. (1976). The aClive metabolite carl>amazepine.IO, II-epoxide may. however, compen· sate for low serum concentrations of the parent dru~
Carbamazepine toxicity. in panicular ataxia, dizziness, diplopia and drowsi ness. is most fre· quent when staning the drug or increasing the dose (Callaghan et al. 1978; Gamstorp 1975). These side eITecls can ~ minimised by gradually increasing the dose. Relatively high serum concentrations may be achieved during the fint 2 days of therapy, al· though acute toxicity can occur when concentra· tions are quite low. Other side effects have been shown to correlate wilh serum concentrations of carbamazepine (Chadwick el al. 1984; Durelli et al. 1985; O'Doughertyet al. 1987; Riva et a!. 1985) or of the metabolite carbamazeplne·IO.II-epoxide (Patsalos et al. 1985; Schoeman et al. 1984). Sev. eral studies have suggested that circadian flu ctua· tions in serum carbamazepine concentration may be responsible for side effects (Bien now 1983; Hoppener et al. 1980; Tomson 1984); toxicity appears to be more likely with serum concentrations above 8 mg/L (O'Dougherty et al. 1987; Riva ct al. 1985). Convulsions secondary to carbamazepine have been reponed (usually in patients rea:iving more than I anllconvulsant) with serum carbamazepine con· centrations ranging from 2 to 12 mg/L (Shields & Saslow 1983; Snead &. Hosey 1985).
7.2 Clonazepam There is considerable interindividual variation in the plasma concentration of clonaupam reo quired for seizure control. An 10itial study of 10
Thernpcuue Monllonng of A nlloon~' ulsan I S
patients suggested a rclativel~ wide therapeutic range of 13 to 72 ~g/L (Drelfuss et al. 1975), whereas a laler stud y reported control in 57 previously untreated patients with plasma clonazepam concent rations ranging from 3 to 42 $Ig/L (Naito el al. 1987). SC"lZure conlro1 was reported In 7 of8 neonates with plasma concentrallons ranging from ::!8 to 1,17 $Ig/l (Andre et al. 1986). Naito et al. (1987) reported Ihat the side effects of clonazepam appear to be dose dependent: howe\·er. this group did not comment on the relationship between plasma concentrations and toxicity. 7.3
Ethosu~mlide
Ethosuximide IS the 'first-line' drug for pew mal Two ~ludl~ have shm'- n the value of monitOring serum concentrations of this agent. and control was usually achle\'ed with figures of be· tween 30 and 100 mg/l (Bro .... ne etal. 1975: SherWin el al. 1973). A later s l ud~ sho .... ed that::! of II children were adequately controlled wit h serum concentrations of 25 and 26 mg/l. respecllvel) (Blomquist & Zellerlund 1985). Further studies arc reqUIred to detenmne whether there IS a lo.... er limn for a therapeutic range for ethosux imide. er,.lep~)
7.4 Phenobarbital As for ot her anticonvulsants. there is conSld· crable tntennd lvidual varia lion tn the serum concentration of phenobarbital required for seizurc conlrol. A minimum concentrallon of 4 mg/L has been suggested (Buchlhal & S\ensmark 1971). based on a study of J3 hospitalised patien ts..... hlle Buchanan & Allcn (1971) reported 69 patients .... lth completc seizure control and serum concentrallons rangtng from 0 to 47 mg/ L (mean 10 mg/l). Ad\'erse ncurologlcal effects (n~stagmus. ataxia and lethargy) arc usuall~ onl~ seen .... nh concen trations greater than 20 mg/l (Gallagher CI al. 1973). whereas many other Side effects (including beha\ioural problems) arc unrelated to SelY e!TeetS from Iht clmldian nuctuallOns of carbamaupone plasma levels. " ell Pacdiatnca Scandlna vlca 72: 397-401 . 1983 BlomqUIst HK , Zetlertund B. EvaluatIon oftn:atment In tYPIcal abscl1
Therapeutic Drug Monitoring of Anticonvulsants
State of the Art
Imtiaz A. Choonara and Anders Rane Insti tute of Child Health, Alder Hey Children's Hospital. Liverpool. England, and Division of Oinkal Phannacology, UniversilY Hospital, Uppsala, S.....eden
Contents
Summary .......................................... . .................. 318 I . Efficacy ................................ .3 19 L I The Lower Limit of the Thel'llpeuhc Ranse ..... 319 L.2 The Upper Limit of the Therapeutic Ranse ..................... 320 2. Toxicity ............................................................. ... 320 3. Indications for Therapeulic DruB MonilorinB of Anticonvulsants ... ...................... 321 4. SJ)«ial Oinical Groups ................. ............................. 321 4. 1 Slatus Epilepticu5 ................................................ . .32 1 4.2 Prq.nancy ................ ................ . ........ 322 4.3 Neonates ..................... .. . 322 4.4 Children ................ ...... 323 5. CompliallCe .......................... . ........ 323 6. Saliva ..................................................................................... . ......... 323 7. Individual Anticon vulsants 324 7. 1 Caroomaupine .................. ... 324 7.2 Clonuepam ............................ ................. . .... 324 7.3 Elh{)$uximide ................. 325 7.4 Phenobarbital ... 325 7.S Phenytoin ... 325 7.6 Valproic Acid .................. 326
There is considcrable interindividual variation in the rela lionship betwttn conlro] of seizures and the serum anticonvulsant concenlra lion. The minimum efTtttive serum conce ntration is dependent on the type and severity of th e epilepsy. and varies from patient to patient. The therapeutic range should be used as a guide to adjust th e dose in ordcr to furthe r improve seizure control or reduce toxicity; the laller is more likel y with higher serum concentrations, but can also be prese nt whcn concenlrations are low. A request for the serum conccntralion of an anticonvulsant should be made only for Bood clinkal reasons, and an interpretation of tha I concentration can only be made if all the relevant clinical detai ls are available.
Therapt'ulle Monl1orlng of AnlleOnl ul!.3nts
319
Indications for [he measurement of serum antleOn, ulsant concentrations Include poor seiZUre' control. tOAICII). suspected gross noncompliance. status epllepllcus and the elapse of 2 to " w~ks after the InlII3110n of themp} . Addillonal drug Iherap). prei'lanc) or illness may alter drug dispoSition in a ",ell controlled pallent and therapeutic drug mOnltormg may. there'fore. help to prevent seiZUre'S secondary 10 these changes. The measure'menl of anllconlulsants 10 sahl'a as opposed to serum may be of benefil In some patients.
It IS almost 30 )ears smce Buchthal and cO",orkcfS studied the relatlonshlp ocI"een serum phen) tOIn concentrallons and control of grand mal epllcps} (Buchthal ct a1. 1960). Since then the conccpt of a therapeutic range has ocen extended and questIOned for dlIYerent anllcoO\ulsants and diagnoses. The abllit} to measurt' plasma concentrations ofanllcOO\ ulsants has ocen shown to help In the control of eplicps) (Lund 197-t North et al. 1980: She,""1n et ai. 1973; Wmg & DulY 1989). It has also been shown that the uncnllcal interpretatIOn of routme $Crum anticonvulsant concentrations docs not Improve patu:nt control (Beardsley el ai. 1983: Froscher et ai. 1981; Woo et al. 1988). There IS no therapeu tic range for any anliconl' ulsan t "'hlch IS applicable to all patients. Even the term 'therapeutic range' itself has been quesllonl'd. and SCI eral authors hale suggested 'optlmal range' as an altematlle (Bartels 1980; Re) nolds et al. 1981: Vajda & Alcardl J 983). There arc 2 maIO conSlderauons ",hl'n a palll'nt IS trcatl'd:
oUlpatil'nts "'lIh grand mal epilepsy. control was beSl achleled wllh serum phen}tom concentratIons abole 15 mg,lL. In 86 pallents (no clinical details 8Jvcn). serum phenytOin concentrations werl' below 14 mg/L and none complained of Side dfects. In contrast. 34 patients had concentrations greater than 30 mgJL: of these 17 (50%) had scvere. and 8 (24%) mild. Side effects. The concept of the therapeutiC range (10 to 30 mgJL for phenytom) had been established. Subsequent studies reponed phen) tOIn toXtCII) at concentrations ab()\e 20 mg,l L (Kull et al. 1964) and therefore the therapcuttc range was revised to 10 to 20 msiL. Therapeutic ranges have Slncc been reported for most anticonvulsants. The alll1 of thiS review IS to dISCUSS some of the problems of the concept of a therapeutIc range.
J. Efficacy: can the patient's disease or
Imllal studies WIth phenytOin Invoh'ed pattents wllh $Clerc grand mal epilepsy. Many pattents can oc adequa te!) con trolled with 'subtherapeullc' plasma concentrations. Feldman and Pippenger (1976) reponed 25 pallents who had been seizurefree for at least 2 yea~: 8 of II patienlS treated wllh phen)toln alone had serum concentrations below 10 msiL (range 0.9 to 6.4 mSlL); 7 of 14 pattents treated "11h phen)toln and phenobarbital In combination had scrum concentratIOns of both anticonl'ulsants oclo" 10 mg/L (range 0 to 9.0 mgj L). Scleral other autho~ have reponed that In some pattents control can be achIeved With ·subthera· peUIiC' concentratIOns of pheny tOin (Dawson & JamIeson 1971 ; Turnbull et al. 1984): Ihis IS con·
s~mptoms
oc controlled? 2. TOXICity: can Side effects of the treatment oc mlnlmlscd or prevented? The delelopment of an anal}lIcal assa} for the measurement ofphen}toin m serum led 10 the subsequent clinical studIes of the relationship between the efficacy and toxicit} ofphen}toln and Its concentratIOn m serum. The mitial prospective slud) of J 2 hospllaltsed patients With severe grand mal eplicps} rcrelvmg phenytoin and phenobarbllal showed conlrol on ly when thl' serum phenytom concentratIOn was greater than 10 mg/L (Buchthal et al. 1960). In the same paper II ",as noted that.
In
I. Eff icacy I.l The Lo"er Llm1l of the Therapeutic Range
elm. pnarmacoklllet. 18 (4) 1990
320
5i5ten! with the clinical impression of many practitioners treating palients with epilepsy. One paper describing the use of ethosuximide gave a suggested therapeutic range of40 to 100 mg,! L. bUI reported I patient with a 93% reduction in seizure frequency with a plasma ethosuximide con~nt ralion
of 16.6 mg/L (Browne et al. 1975). A
prospective study of previously untreated patients with epilepsy showed thaI, in many patients. complete control of seizures could be achieved with 'subtherapeul ic' serum concentrations of phenytoin (Callaghan el al. 1985). A recent randornised. controlled trial showed that there is no advantage in increasing the dose of phenytoin or phenobarbital in patients with 'subtherapeutic' serum concentrations who have been seizure free for 3 months or more (Woo et al. 1988). It is clear. therefore. that no universal lower limit exists for the therapeutic range ofanticon vu lsants. Due consideration must be given to the patient's history offits before the drug dose is increased in patients with apparently subtherapeutic drug concentrations.
1.2 The Upper Limit of the Therapeutic Range Although the initial repons involved patients with severe grand mal epilepsy, subsequent studies have shown that some patients require serum concentrations above the upper limit of the therapeutic range. One long term prospt(:tive study showed considerable interindividual variat ion in the therapeutic serum phenytoin concentration for each patient (lund 1974): several patients with severe grand mal epilepsy required serum concentrations above 20 mg/l to achieve satisfactory control. Although most prospective studies of efficacy and serum concentrations involved patients with grand mal seizures, some included individuals with partial seizures. In some studies. no correlation was found between the serum concen trations ofphenytoin (Gannaway & Mawer 1981: Turnbull et al. 1984), carbamazepine (Strandjord & Johannessen 1980) or valproic acid (Turnbull et at. 1983) and the control of partial seizures. A prospt(:tive study of 40 patients with grand mal epilepsy and 38 with
partial seizures with or without grand mal epilepsy showed that higher plasma concentrations of phenytoin, phenobarbital and carbamazepine were needed for complete control in those with partial seizures (Schmidt et at. 1986). The upper limit should. therefore. be considered as a guide. If clinically indicated it may be necessary to exceed it. It is important to be aware. however. that seizures can be induced by anticonvulsants: this phenomenon has been reported following therapy with phenytoin (Levy & Fenichel 1965). mephenytoin (Troupin & Ojemann 1975), carbamazepine (Shields & Saslow 1983: Snead & Hosey 1985) and diazepam (Prior et a!. 1972). With phenytoin and mephenytoin, these cases are usually (but not always) associated with a high serum drug concentration (Stilman & Masdeu 198 5: Troupin & Ojemann 1975). Therefore, in a poorly controlled patient with a high serum phenytoin concentration. the elinician should consider both an increase and a decrease in the dose as therapeutic possibilities.
1. Toxicity The original studies with phenytoin (Buchthal et al. 1960: Kutl et al. 1964) suggesled a clear relationship between serum concentration and toxicity. Buchthal and co-workers reported no side effccts in patients with serum phenytoin concentrations below 14 mg/l. and severe toxicity was present only with concentrations greater than 30 mg/L. Kutt and co-workers reported nystagmus with seru m phenytoin concentrations of 15 10 25 mg/l. ataxia with concentrations of 30 to 45 mg/ L and mental slowing (confusion, disorientation and lethargy) with concentrations above 40 mg/L. Subsequent studies supported these findings(Haerer & Grace J 969) and free phenytoin concentrations were shown to have an even closer correlation with toxici ty (Booker & Darcey 1973). Side effects of anticonvulsants aTe common (Choonara 1988a) and the reported incidence depends on the thoroughness of the $Carch for them (Herranz et al. 1982). The original studies of phenytoin toxicity (Buchthal et al. 1960) did not
TherapeullC Momlonng or Antlcon\ ulsanl~
tf) to dr/eet side effects. bu t relied on those that the pallent complamed or. Moreover. gi ngival hypenroph) was specificall) e"dudcd. T he more subtlt' eflects ofantlconvulsants on cogni tive function (Thompson & Trimblc (982) were nOt realised at the IImc. SIde effects arc dose dependent. allergic or Id io. s)ncrnllc. \\ hcrcas one "ould C,\pect a correlatIOn bet"ecn to'ICII) and serum concentrallon for thc first menlloned. the last 2 are unrelated to serum concentration. It IS therefore not surpnSlng that somc groups ha\e reponed a correlation octwttn serum concentration and some adverse drug react ions. e.g. ski n reactions (ChadwIck et al. 1984; Wilson et al. 1978) and cardiotoxlci ty (Durelll ct al. (985). "hcreas others have found none (Choonara 1988b; Gram et al. 1979; Herranz et al. 1982). Side effects ma~ also be related to the rormatlon ofa mNaboh te (Park & Kittenngham 1987) "hich has been proposed for carbamazepme-IO.II-epo,Ide (Patsalos et a1. 1985: Schoeman et al. 19!{4). J ust as the ther::lpcutlc rangc must be considered as a general gUidc fo r cflicac~. so II must be realised that to"lelt) can occur with subtherapeutlc or therapeutic concentrallons. and that some patients can tolerate coneenualions above the therapeullc range without an) SIde effects.
3. In dications for Th erapeutic Drug Monitoring of Anticonl'uisants The points abo\e arc mentioned as an argumenl for the appropnate usc of therapeutIc monltormg faclh\les, and to emphasise thc problems of a dogmatIc attitude towards serum concentration anal)scs. There must be a chnlcal rcason to rcqucst a scrum concentration of an antlcon\UlsanL routine monltonng of anllcon, ulsants on all outpallents has no ,alue. eSpt'c13l1) If the patient's sclzures arc "ell controlled. The follOWing arc the most acceptable reasons for requesting the scrum concentratIOn anal)sls of an antlcon, ulsant: I. Poor sclzure Control. 2. TOXlcll) . 3. Suspected gross noncompliance.
321
4. Status epilepllcus. 5. At 2 to 4weeks after Initiation o f therap y. 6. Modification of concomitant drug therapy (additions. dose changes. discon tinuation). 7. Pregnancy. 8. Itlness (fever. hepatIC or renal malfunction). 9. Research. ThiS an lde does not deal wllh drug In teractions between ant lconvulsants as the au thors are unaware of a n) cll mcal Indication fo r treatment with more than I anticonvulsan t at a time; on the contrary. there h ave been a num ber of re ports which have shown that I drug alone is superio r t o comblOallon therap} In terms of effieac) (Lund 1974: Reynolds & Shorvon 1974). Most studies have tn· vol'OO collectIOn of trough samples. i.e. prior to the adm lOlstra tlon of a dose: this ISpreferable. bu t often ImpractIcal tn clinical Sllua tlons. It IS there· fo re essenual to know the lime of admtn tstrat lon of the last dose and thiS should be noted on the analYSIS request form . It takes a period equI\alent to 5 half-Ines before steady-state concentratIons arc reached and tn the case of carbamazeptne. beea usc of ItS au tolnductlon of Its own meta bol ism. Itta kes even longer. The lime reiallonshi p between an cffeclne serum concentration and efficacy has not been established e'cept tn cases of status epilepticus. and the authors therefore "ould recommend a minImum of7to 14 days' admtnistration of a ny anticonvulsan t before the serum eoncentrallon is checked (unless the case IS one of sta tus eptleptlcus). In the casc of phenobarbital it rna)' take up to 4 weeks for steady-state levels to be achie"ed because of the long half-life of that drug.
4. Special Clinical Groups 4.1 Status Epilepllcus Intra,enous d13zepam or clonazepam are the drugs of chOIce for acute SClZures. They ar(: gl\en as a bolus and rna) be conllnued as an InfUSIOn. .,,),5 both cfficac) (termlnallon of the SClzu res) and tOXIClt} (respirator) depression) are Immediate. there IS no need \0 measure theIr serum coneentrallons In the short term. Ifa benzodlazeptne fa ds.
Om Phurmuro/..m l'l 18 (4) /990
322
ph~nytoin is often used: a loading dose of ph~ny toin 15 mg/kg administered as an Infusion la, a ra te of 40 to 50 mg/min in adults or I to 3 mg/ kg/min in children (Albani 1977») in patients not previously rea:iving the drug should result in plasma phenytoin conct'ntrations above 10 mg/L within 10 minutes (Salem et al. 1981). The response to phenytoin in status epilepticus appears to b(: related to the underlying disease. and hen~ widely varying response may b(: seen in different patient groups. Two separate studies have shown a similar serum phenytoin concen tration in responders and nonresponders (Cranford et a\. 1979; Wallis et a!. 1968). Because of the high morbidity and mortality associated with status epilepticus. it is worthwhile to check the serum phenytoin concentration within 30 minutes of the end of the infusion to ensure the concentration is at least 15 to 20 mg/L (Wallis et a!. 1968). If the patient continues to have seizures despite such a serum concentration. it is probably advisable to try another anticonvulsant rather than to con tinue with further doses of phenytoin . Paraldehyde and valproic acid are also effecti ve in status epilepticus. but there is insufficien t data to indicate the need for measurement of serum con~ntrations of these drugs.
4.2 Pregnancy Many women with epilepsy experience an increase in seizure frequency during pregnancy (Knight & Rhind 1975; Schmidt et a1. 1983). Numerous physiological changes occur during pregnancy: the metabolic clearance of drugs appears to be enhanced (Dam et al. 1979; HOgstedt et al. 1983), and there is also an increase in the unbound fraclion of drug (Chen et al. 1982). Despite this increase. the free serum concentrallon of phenytoin IS decreased in late pregnancy (Chen et al. 1982). It has been suggested that monitoring of saliva phenytoin concentrations be carried ou t during pregnancy and in the postnatal period (KnOll et al. 1986). as these correlate with the free plasma concentration. Further prospective studies are required to confirm this suggcstion. In the case of
carbamazeplne there is a small increase In the unbound fracllon and ils metabolite carbamazeplne. 10, II -epoxide. and an increase in the ralio ofepox. ide to carbamazepine (Yerby et al. 1985). suggesting a higher bioactivation rate in prqnancy. Even though pregnancy may alter seizure thresholds as well as drug disposition. and the ther· apeutic range during pregnancy has not 1>«n d~ tennined. until such studies han 1>«n perfonned the aim should be to keep anticonvulsant concentrations at the upper end of the therapeutic range or al the level that has controlled each individual patient prior to the pregnancy. There is greater risk associated with insufficient seizure con trol than of a malformation of the child occurring as a result of treatment. 4.3 Neonates Phenobarbital has 1>«n the drug ofchol~ In the neonatal period. &vcral studies suggest thai control can be ach ieved wi th serum concentrations of 15to 30 mg/l (Pippenger & Rosen 197.5). 12 to]O mg/l (Jailing 19 75) and 17 to 25 mg/l (Lockman et al. 1979). Gal and co-workers used a loading dose of 15 mg/kg followed by additional doses of 5 to 10 mg/kg every 5 to 10 minutes until control was achieved (maximum dose 40 mg/kg) in a group of 71 neonates (Gal el al. 1982). They controlled seizures in 60 (85%) of the neonates. of whom ]6 (60%) responded to plasma conctn trations below 20 mg/ land 17 (28%) 10 concentrations of 20 to 30 mg/ L. Seven neonates ( 12%) responded on ly to plasma concen trations above 30 mg/L. A recent study shows that there is little value in funher doses of phenobarbital once a concentration of 40 mg/l is achieved (Gilman et al. 1989). Sedation. poor fceding and hypotonia have be(:n reported at conctntrations above 50 mg/l (Gilman et al. 1989).60 mg/l (Pippenger & Rosen 1975) or 70 mg/l (Painter et al. 1978). Studies of the pharmacokinetics of phenytoin and valproic acid in the neonatal period havc also been carried out. but there is insufficient data to give a therapeutic range in this agc group.
Therapeutic Monllonng of AnliconvulsaniS
4.4 Children Many of the initial studies Involved children as well as adult patients. and several groups have looked specifically at children. Borofsky and coworkers (1972) reported control in children of all ages with serum phen)toin concentra tions of I to 18 mglL Drug to'\ICIt) (gross ataxia) was reponed in 8 of the 53 childrcn with serum phenytoin concenlraBons ranging from 17.8 to 48 mg/L The same researchers late; subdivided these 53 children into 10 who showed a d,rcct correlallon between seIZure control and serum concentrallon. 8 who failed to achle\e control despJle a high serum concenlration and 30 who were controlled on thc Imtlal dose of phenytOin (Borofsky et al. 1973). Dawson and Jamieson (1971) showed that monitonng serum phenytoin concentralions and subsequently in_ creasing the dose led to improved seizure control in II of 30 children. In a study of 109 children (Norell et al. 1975), serum phenytoin concentrations in those who had been seizure free for a minimum of 3 months ranged from I to 28 mglL In 9 children seizures were only controlled when serum phenytoin concentrations wcrc above 12 mgl L These studies suggest that there is as much interindividual variation in Ihc therapeutic serum concentration ofphcll) tOIn In children as In adults. The availabllll) of modern telemetry facII,tlcs has permlttcd the obJcctlve assessment of EEG·venfied seizures In relation to the serum antIconvulsant conccntratlOn (Braathen et al. 1988: St;\hl et al. 1983). There have been many studies of the use of phenobarbital. valproic acid and other anticonvulsan ls in chi ldren with febri le seizures. although it has been suggested that there is little justification for using these potentially toxic drugs for a condition with such a good long term prognosis when rectal diazepam is available [see the rCVlew by Knudsen (1988)].
5. Complianu Although the measurement of serum an IIconvulsant concentrations can assist In detecting gross noncompliance (concentrations below thc limll of
323
detection suggest that the patient is not laking the medication), other factors may also explain low serum concentrations: examples are low dose. increased metabolism or malabsorption. Carefully worded questions. for example ' How often do you forget to take rour medicineT. 'At what time of the day are you most likely to forget?'. which do not imply criticism. should be used 10 detecting noncompliance (Mucklow & DolleT)' 1978: Pearson 1982). Such quesllons are more uscful than routine monitoring of serum anticonvulsant concentrations In all outpatients. In a prospective study of 96 children. 18 gave a history of noncompliance whereas rou!Jne measurement of serum concentratIOns dctected gross noncompli. ance III on I) 3 (2 of whom gave a positive history of noncompliance). and r::llsed the POSSibI lity In 4 (3 of whom sa"e a positive history of noncompli_ ance) [Choonara 1988b]. Measurement of serum anticonvulsant concentrations to check for noncompliance should only be considered in patients with poor control who den) it despite careful questioning.
6. SaJil'a Saliva is a low protein fluid. and saliva drug concentrations therefore reflect Ihe amount of un· bound drug. This r('lallonshlp has been documented for carbamazepine. phenyt01l1. phenobar. b,tal and pnmldone (McAultfTc Ct al. 1977). The protein b1l1ding of ethosuximldc is insignificant. and therefore salivary and serum concentrations are considered to be the same. The pKa of phen. obarbital is similar to the pH of saliva. and it has been suggested that pH be measured simultan· eously (McAul1ife ct a1. 1977). but this has been disputed b) others (Fnedman et al. 1981: Tokugawa et al. 1986). Valproie acid has a low distribution into sall\ a. and thus measurement of salivaT) concentrations IS of no value (Fung & Ueda 1982). However. the collcction of saliva does not In\ohe \enepunc(urc. and thiS may be of value in some children. Most studIes have correlated salivary concentratIOns wllh serum concentrations without regard
Clm Pharmaroklnf'f 18 (4)
"""
,-, ""'"
CarblllNlz. . . .
~"
.
1. 13
23
1).720
"
CIonaz~m
EthOSlI~lmide
Pt\enobaJb;t.,
-"~
,..,.
'&'101 3·22 ' , '5 '.18
I b c
No'
"
3S
" "
""
R.ler--=StrandfOtd &. JohannesMII (1980) Cahghan.l .,. (1978) Dr~tuss .1 ., (1975) N.i1o" aI . ('987) BrOWN.' II. (1915) 8ucnltlal &. SVlln,m.rtI ('971) FeICIlNln &. Pippenglf
(1976)
Sorolsky lit .,.
(1972)
Number 0 1 controlled patient, In study
,gIL_ Not II receiylI'Ig monolheriPY
for efficacy or 10)!:iciIY. Rylance and Moreland ~tween saliva carbamazepine and phen ytoin conCtntrations and toxicity, but no correlation with efficacy. They sug· gested a therapeutic range for carbamazepine in sa· liva of 1.2 to 3.5 mg/L and for phenytoin ~twttn 5 and 25 mg/L. Nevenheless. the value of moni· toring drug conCtntralion in saliva in the clinical situalion has not been fully assessed, and fun her studies are required.
(1981) found a posilive correlation
7. Individual .4nliconvulsants Table I shows therapeutic ranges for patien ts reo ceiving monotherapy with the drugs reviewed ~Iow.
7.1 Carbamazepine Carbamazepine is the drug of choice for partial com pie)!: seizures. grand mal and temporal lo~ epilepsy. It is as effective as phenytoin butlcss toxic (Callaghan et al. 1985; Cereghino et al. 1974; Shor· von el al. 1978: Troupin et al. 1977). Initial studies suggested a therapeutic range above 4 mg/l (Dam
f99{)
et al. 1975) and between J and 12 mg/L (Strand· jord &. Johannessen 1980). There is. howe\·er. con· siderable interindividual vanatlon in the serum concentration of carbamaztpme reqUired to achie\'e control. In 2 separate prospectIVe studies. complete control was achieved in 5 of25 pat,ents with serum concentrations below 4 mg/l (Shorvon et al. 1978) and in 13 of 28 patients with concentrations of I to 8 mg/l (Callaghan et al. 1978). Childrcn with low serum concentrations of carba mazepine who are nevertheless well controlled have been reponed by Rane et al. (1976). The aClive metabolite carl>amazepine.IO, II-epoxide may. however, compen· sate for low serum concentrations of the parent dru~
Carbamazepine toxicity. in panicular ataxia, dizziness, diplopia and drowsi ness. is most fre· quent when staning the drug or increasing the dose (Callaghan et al. 1978; Gamstorp 1975). These side eITecls can ~ minimised by gradually increasing the dose. Relatively high serum concentrations may be achieved during the fint 2 days of therapy, al· though acute toxicity can occur when concentra· tions are quite low. Other side effects have been shown to correlate wilh serum concentrations of carbamazepine (Chadwick el al. 1984; Durelli et al. 1985; O'Doughertyet al. 1987; Riva et a!. 1985) or of the metabolite carbamazeplne·IO.II-epoxide (Patsalos et al. 1985; Schoeman et al. 1984). Sev. eral studies have suggested that circadian flu ctua· tions in serum carbamazepine concentration may be responsible for side effects (Bien now 1983; Hoppener et al. 1980; Tomson 1984); toxicity appears to be more likely with serum concentrations above 8 mg/L (O'Dougherty et al. 1987; Riva ct al. 1985). Convulsions secondary to carbamazepine have been reponed (usually in patients rea:iving more than I anllconvulsant) with serum carbamazepine con· centrations ranging from 2 to 12 mg/L (Shields & Saslow 1983; Snead &. Hosey 1985).
7.2 Clonazepam There is considerable interindividual variation in the plasma concentration of clonaupam reo quired for seizure control. An 10itial study of 10
Thernpcuue Monllonng of A nlloon~' ulsan I S
patients suggested a rclativel~ wide therapeutic range of 13 to 72 ~g/L (Drelfuss et al. 1975), whereas a laler stud y reported control in 57 previously untreated patients with plasma clonazepam concent rations ranging from 3 to 42 $Ig/L (Naito el al. 1987). SC"lZure conlro1 was reported In 7 of8 neonates with plasma concentrallons ranging from ::!8 to 1,17 $Ig/l (Andre et al. 1986). Naito et al. (1987) reported Ihat the side effects of clonazepam appear to be dose dependent: howe\·er. this group did not comment on the relationship between plasma concentrations and toxicity. 7.3
Ethosu~mlide
Ethosuximide IS the 'first-line' drug for pew mal Two ~ludl~ have shm'- n the value of monitOring serum concentrations of this agent. and control was usually achle\'ed with figures of be· tween 30 and 100 mg/l (Bro .... ne etal. 1975: SherWin el al. 1973). A later s l ud~ sho .... ed that::! of II children were adequately controlled wit h serum concentrations of 25 and 26 mg/l. respecllvel) (Blomquist & Zellerlund 1985). Further studies arc reqUIred to detenmne whether there IS a lo.... er limn for a therapeutic range for ethosux imide. er,.lep~)
7.4 Phenobarbital As for ot her anticonvulsants. there is conSld· crable tntennd lvidual varia lion tn the serum concentration of phenobarbital required for seizurc conlrol. A minimum concentrallon of 4 mg/L has been suggested (Buchlhal & S\ensmark 1971). based on a study of J3 hospitalised patien ts..... hlle Buchanan & Allcn (1971) reported 69 patients .... lth completc seizure control and serum concentrallons rangtng from 0 to 47 mg/ L (mean 10 mg/l). Ad\'erse ncurologlcal effects (n~stagmus. ataxia and lethargy) arc usuall~ onl~ seen .... nh concen trations greater than 20 mg/l (Gallagher CI al. 1973). whereas many other Side effects (including beha\ioural problems) arc unrelated to SelY e!TeetS from Iht clmldian nuctuallOns of carbamaupone plasma levels. " ell Pacdiatnca Scandlna vlca 72: 397-401 . 1983 BlomqUIst HK , Zetlertund B. EvaluatIon oftn:atment In tYPIcal abscl1
