PATHOGENESIS, PREVENTION OF RECURRENCES AND OUTCOME OF FEBRILE SEIZURES

RITATARKKA

Department of Paediatrics,University of Oulu

OULU 2003

RITA TARKKA

PATHOGENESIS, PREVENTION OF RECURRENCES AND OUTCOME OF FEBRILE SEIZURES

Academic Dissertation to be presented with the assent ofthe Faculty of Medicine, University of Oulu, for publicdiscussion in the Auditorium 12 of the University Hospitalof Oulu, on September 5th, 2003, at 12 noon.

OULUN YLIOPISTO, OULU 2003

Copyright © 2003University of Oulu, 2003

Supervised byDocent Heikki RantalaProfessor Matti Uhari

Reviewed byProfessor Lennart von WendtProfessor Tuula Äärimaa

ISBN 951-42-7088-6 (URL: http://herkules.oulu.fi/isbn9514270886/)

ALSO AVAILABLE IN PRINTED FORMATActa Univ. Oul. D 741, 2003ISBN 951-42-7087-8ISSN 0355-3221 (URL: http://herkules.oulu.fi/issn03553221/)

OULU UNIVERSITY PRESSOULU 2003

Tarkka, Rita, Pathogenesis, prevention of recurrences and outcome of febrile seizures Department of Paediatrics, University of Oulu, P.O.Box 5000, FIN-90014 University of Oulu,Finland Oulu, Finland2003

Abstract

Febrile seizures (FS) occur in 2–5% of children. Their pathogenesis is unknown. Elevated levels ofprostaglandins (PG) have been found in cerebrospinal fluid after such seizures, and a third of allpatients have recurrences. No safe ways of reducing the risk of recurrences have been found. Theoutcome has been shown in prospective studies to be good, by they have been linked to mesialtemporal sclerosis (MTS) in patients with severe temporal lobe epilepsy (TLE).

The aim was to analyze the records on the role of PGs in the pathogenesis of FS, to find risk factorsfor recurrences that are amenable to intervention and to evaluate the prevention of recurrences andthe connection of FSs with MTS.

We performed a systematic review of the effect of PGs and their synthetase inhibitors on seizuresand a meta-analysis of the prevention of recurrences. The prophylactic effect of diazepam andacetaminophen on recurrences was evaluated in a placebo-controlled trial with 180 FS patients, andrisk factors for recurrences were analysed from these data. To find MTS, MRI volumetry wasperformed after 12 years of follow-up on 64 cases chosen out of 329 unselected FS patients: twenty-four with a prolonged initial seizure, eight with a later unprovoked seizure and 32 age, sex andhandedness-matched controls.

PGD2, PGE1 and PGE2 had mainly anticonvulsive effects and PGF2alfa proconvulsive ones.NSAIDs had seizure-modulating effects in adult animals ranging from attenuation to provocation.Each degree of increase in fever doubled the recurrence risk, and each febrile episode increased it by18%. The meta-analysis showed phenobarbital and valproate to prevent recurrences, but they cannotbe recommended for FS as they have severe side-effects. The meta-analysis nullified the allegedeffect of diazepam, and neither this nor acetaminophen prevented recurrences in a clinical trial. NoMTS was found in any patient group.

PGs may be involved in the pathogenesis of FS. No safe prophylaxis for recurrences is available,although the effect of antipyretics needs further evaluation. Measures to reduce feverish infections inorder to prevent FS recurrences seem logical. MTS is uncommon even after prolonged FS.

Keywords: acetaminophen, antiepileptic drugs, antipyretic agent, diazepam, febrileseizure, mesial temporal sclerosis, prostaglandin

To Tatu and Antti

AcknowledgementsThe work for this thesis was carried out at the Department of Paediatrics, University ofOulu, during the years 1992-2003.

I thank Professor Mikko Hallman, MD, Head of the Department, for providing theoptimal facilities for successful research at the Department. I express my deepestgratitude to my supervisors Docent Heikki Rantala, MD, and Professor Matti Uhari, MD,for the excellent guidance and enormous patience during the decade of my Paediatricresearch.

I am grateful to the official referees Professor Tuula Äärimaa and Professor Lennartvon Wendt for valuable and constructive comments on the manuscript.

My warmest thanks are due to Virpi Räikkä, RN, for friendship, everlasting optimismand essential help especially in the fifth study of this thesis. I also wish to thank my co-authors Leena Vainionpää, MD, Tytti Pokka, BSc, and Eija Pääkkö, MD, with whom itwas a great pleasure to work, and Malcolm Hicks, MA, for revising the English languageof this thesis.

I am sincerely grateful to all the children and their parents who made this workpossible.

I wish to express my loving thanks to my mother Mirva Kurttila for providing constantsupport and advice in uncountable occasions during the writing of this synthesis, and myfather Eero Kurttila for frequently helping my family during the hardest times in everydaylife, and my brothers Juri and Jarkko for honest opinions, and Tanja, Visa, Sanni and AlttiKurttila for relaxing moments in between the research work while visiting thegrandparents in Oulu. The warmest thanks also belong to my mother- and father-in-lawManna and Matti Tarkka and sisters-in-law Elina, Helena and Emma for all help andsupport during these years. I wish to thank my friend Anina Raitio for sharing the joysand sorrows confronted in research work. My husband Tatu and son Antti are everythingto me and bring the most precious love and happiness into my life. I cannot find wordsenough to describe the gratitude I owe them just for being there.

This work has been supported by the Alma and K.A. Snellman Foundation, Oulu,Finland, the Arvo and Lea Ylppö Foundation, the Research Foundation of the OrionCorporation, the Foundation for Paediatric Research in Finland.

Abbreviations95% CI 95% confidence interval COX cyclo-oxygenase CSF cerebrospinal fluidEEG electroencephalogramFS febrile seizureHHV-6 human herpes virus-6IL interleukin MES maximal electroshockMRI magnetic resonance imagingMTS mesial temporal sclerosisPG prostaglandin OR odds ratioPTZ pentylenetetrazoleRR risk ratioSPGR spoiled gradient echoSPSS Statistical Package for Social SciencesT1 longitudinal relaxation timeT2 transverse relaxation timeTLE temporal lobe epilepsyTNFalfa tumour necrosis factor alpha

List of original publicationsI Rantala H, Tarkka R & Uhari M (2001) Systematic review of the role of

prostaglandins and their synthetase inhibitors with respect to febrile seizures.Epilepsy Research 46: 251-257.

II Tarkka R, Rantala H & Uhari M (1998) Risk of recurrences and outcome after thefirst febrile seizure. Pediatr Neurol 18: 218-220.

III Rantala H, Tarkka R & Uhari M (1997) A meta-analytic review of the preventivetreatment of recurrences of febrile seizures. J Pediatr 131: 922-925.

IV Uhari M, Rantala H, Vainionpää L & Kurttila (Tarkka) R (1995) Effect ofacetaminophen and of low intermittent doses of diazepam on prevention ofrecurrences of febrile seizures. J Pediatr 126: 991-995.

V Tarkka R, Pääkkö E, Pyhtinen J, Uhari M & Rantala H (2003) Febrile seizures andmesial temporal sclerosis – no association in a long-term follow-up study.Neurology 60: 215-218.

Contents

Abstract Acknowledgements Abbreviations List of original publications 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 Review of the literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.1 Definition of febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.2 Occurrence of febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 Aetiology and pathogenesis of febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.3.1 Family history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.3.2 Fever and febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.3.3 Prostaglandins and cytokines in febrile seizures . . . . . . . . . . . . . . . . . . . 19

2.4 Prevention of recurrences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.5 Outcome of febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.5.1 Mesial temporal sclerosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Aims of the research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Material and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4.1 Structure of the research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.2 Methods for the review of prostaglandins and their synthetase inhibitors

in relation to seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.3 Subjects and methods in the prophylactic treatment trial and the study

of risk factors for recurrences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254.3.1 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254.3.2 Treatment protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254.3.3 Evaluation and follow-up of the patients . . . . . . . . . . . . . . . . . . . . . . . . . 264.3.4 Risk factors for recurrences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.4 Meta-analysis of prophylactic treatment against recurrences . . . . . . . . . . . . . . 274.5 Patients and methods in the mesial temporal sclerosis survey . . . . . . . . . . . . . 284.6 Ethical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.7 Statistical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305.1 Endogenous prostaglandins, prostaglandin synthetase inhibitors and seizures 305.2 Risk factors, prognosis regarding recurrences and results of the treatment trial 33

5.2.1 Occurrence of fever episodes and seizures during follow-up . . . . . . . . . 335.2.2 Risk factors and the prognosis regarding recurrences . . . . . . . . . . . . . . . 335.2.3 Effect of acetaminophen and low dose diazepam on recurrences . . . . . . 33

5.3 Meta-analysis of preventive treatment for febrile seizure recurrences . . . . . . . 355.4 Results of the mesial temporal sclerosis survey . . . . . . . . . . . . . . . . . . . . . . . . 36

6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.1 Study design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.2 Prostaglandins and their synthetase inhibitors in relation

to febrile seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.3 Opportunities to modify risk factors for recurrences . . . . . . . . . . . . . . . . . . . . 406.4 Antipyretic drugs for preventing recurrences . . . . . . . . . . . . . . . . . . . . . . . . . . 406.5 Anticonvulsant treatment for preventing recurrences . . . . . . . . . . . . . . . . . . . . 416.6 Febrile seizures and mesial temporal sclerosis . . . . . . . . . . . . . . . . . . . . . . . . . 42

7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Original publications

1 IntroductionFebrile seizures (FSs) are experienced by 2-5% of all children, and almost a third willsuffer from further recurrences (Nelson & Ellenberg 1978, Verity et al. 1985a). FSs arefrightening for the families, and over 80% of parents have been reported to fear the deathof the child during the initial attack (Baumer et al. 1981). FSs do not have adverse effectson intelligence or learning abilities (Ellenberg & Nelson 1978, Verity et al. 1985b). Therisk of epilepsy is 2-3% in a child experiencing a FS, which is higher than the < 1% riskof epilepsy in children with no FSs (Nelson & Ellenberg 1976, Verity & Golding 1991,Berg & Shinnar 1996).

The pathogenesis of FSs is unknown. The role of prostaglandins (PGs) in feverinitiation is evident, as inhibitors of PG synthesis effectively block the coordinated febrileresponse to cytokine signals such as IL-1 (Saper & Breder 1994). Löscher and Siemes(1988) reported an increase in PGE2 levels in CSF during fever and after FS comparedwith an afebrile state, and the concentrations of PGF2alfa in CSF have been shown to beover fivefold after FS compared with unprovoked seizures (Tamai et al. 1983). Theseizure-modulating effects of endogenous PGs in animals have ranged from inhibition toprovocation (Climax & Sewell 1981, Rosenkranz & Killiam 1981). The PG system in thebrain may be involved in FSs. PG synthetase-inhibiting drugs are used as antipyretics inchildren, but their influence on PG production in the brain and on seizure susceptibility inhumans is unknown.

Prevention of FS recurrences is desirable at least from the parental standpoint. Ayoung age at the initial FS and a family history of FS have been reported to increase therisk of recurrence (Berg et al. 1992), but it is only the finding of risk factors that areamenable to intervention that can provide measures for protecting children from furtherrecurrences.

The effect of continuous anti-epileptic medication, i.e. phenobarbital and valproate, onrecurrences has been demonstrated (Camfield et al. 1980, Mamelle et al. 1984), but theirprophylactic use for preventing FS recurrences is restricted by their side-effects(Committee on Drugs 1982, Farwell et al. 1990). The effect of intermittent diazepam hasnot been clear in clinical trials (Knudsen & Vestermark 1978, Autret et al. 1990, Rosmanet al. 1993). As a high fever increases the risk of FSs (Rantala et al. 1995), the use ofantipyretic drugs has seemed to offer a solution for preventing recurrences, but theireffect has not been properly evaluated and there is no evidence of any prophylactic

16

prevention of recurrences (van Stuijvenberg et al. 1998, American Academy of Pediatrics1999).

Retrospective studies have shown that mesial temporal sclerosis (MTS) occurs moreoften and is more severe among temporal lobe epilepsy (TLE) patients with previous FSsthan in ones with no such history, so that a causal relation has been suggested (Cendes etal. 1993, Trenerry et al. 1993). This is in contradiction to the good clinical outcomereported for FSs (Nelson & Ellenberg 1978, Verity et al. 1985b).

2 Review of the literature

2.1 Definition of febrile seizures

FSs usually occur between 3 months and 5 years of age. They are associated with fever,but without evidence of intracranial infection, a defined cause or previous non-febrileseizures (Consensus Development Conference on Febrile Seizures 1980). Most FSs aresingle generalized seizures of duration less than 15 minutes, but 10–30% are complicated,i.e. prolonged (duration more than 15 minutes), multiple (with a recurrence within 24hours) or having focal features (Nelson & Ellenberg 1978, Verity et al. 1985a, Knudsen1990). FSs have a good prognosis and are to be distinguished from epilepsy, which ischaracterized by recurrent unprovoked seizures (Consensus Development Conference onFebrile Seizures 1980). Since fever can provoke seizures in epileptic patients at any age,an initial seizure occurring during fever can also be the first manifestation of epilepsy, butone seizure with or without fever never justifies a diagnosis of epilepsy.

2.2 Occurrence of febrile seizures

The occurrence of FS in Western Europe and the United States is 2-5% (Nelson &Ellenberg 1978, Verity et al. 1985a, Forsgren et al. 1990b). Nelson and Ellenberg (1978)observed a statistically significant difference between black and white children in thisrespect in a national cohort study (4.25% vs. 3.5%). The occurrence of FSs in Japan hasbeen 7% (Knudsen 1990). Most population-based studies have not found any significanteffect of gender (Nelson & Ellenberg 1978, Verity et al. 1985a, Knudsen 1990). In nearly90% of cases the age at the onset of FSs is less than three years, and half of the patientsexperience their first FS during the second year of life (Verity et al. 1985a).

In a meta-analysis of 14 evaluations of the predictors of recurrences, Berg et al. (1990)reported the average recurrence rate for FSs to be 34.3%, the risk of several recurrencesbeing 16%. Half of the recurrences are reported to occur within six months of the initialFS and three quarters within a year (Nelson & Ellenberg 1978). A first FS at an age of

18

less than one year doubled the risk of recurrences relative to a first attack later on. Youngage has also been found to be the most powerful risk factor for multiple recurrences (Berget al. 1990).

2.3 Aetiology and pathogenesis of febrile seizures

2.3.1 Family history

Berg et al. (1995) reported that children having a FS had an odds ratio (OR) of 6.5 for apositive family history of FS in first-degree relatives and 3.6 in second or higher-degreerelatives compared with children with no history of FSs. A positive family history of FSnearly triples the risk of recurrences, but there is no evident influence of unprovokedseizures in family members on the recurrence rate (Berg et al. 1990). The inheritance ofFS has been thought to be multifactorial and polygenic (Forsgren et al. 1990a).

2.3.2 Fever and febrile seizures

Each additional degree of body temperature above 101°F (38.3°C) almost doubled therisk of FS in the series of Berg et al. (1995) (OR 1.8, p <0.001), and Rantala et al. (1995)likewise reported that children with FS had a higher body temperature both at the time ofhospital admission and during the first three days in hospital than age and sex-matchedcontrols with a feverish infection without seizures (p <0.01).

The lower the body temperature during the initial FS, the greater is the risk ofrecurrences (El-Radhi & Banajeh 1989, Berg et al. 1992, Offringa et al. 1992), so thateach degree of increase in body temperature during the first seizure reduces therecurrence risk by 18% (RR 0.82, 95%CI 0.69-0.97) (Berg et al. 1992).

To evaluate the connection between the duration of fever prior to the initial seizure andrecurrences, Berg et al. (1992) compared recurrences between children with fever lastingless than an hour, from one to 24 hours and more than 24 hours. The risk of recurrencesdecreased as the duration of fever prior to the first FS increased (RR 0.4, 95%CI 0.35-0.66).

FSs mostly occur during fever caused by common viral agents (Lewis et al. 1979,Rantala et al. 1995). Rantala et al. (1995) pointed to two significant differences in theclinical diagnosis between their FS and control groups, the latter having more pneumonia(9% vs. 0%) while the former had more exanthema subitum (9% vs. 0%), and speculatedthat the difference in clinical symptoms between the groups was partly due to patientselection, as the paediatric outpatient clinic mainly treats children referred from primaryhealth care. Exanthema subitum is caused by human herpes virus 6 (HHV-6). Barone etal. (1995) suggested that primary HHV-6 infection might be important in the first FS, andit has also been suspected to increase the risk of recurrences (Kondo et al. 1993). Theseassumptions have not been confirmed in a prospective setting, however. Similarly,

19

patients with a first FS did not have a higher incidence of HHV-6 infection than age-matched controls (Hukin et al. 1998), nor were any differences in FS recurrence ratesobserved when patients with an initial FS during HHV-6 infection were compared tothose presenting during an infection caused by other pathogens (Jee et al. 1998). ThusHHV-6 does not seem to have any special role in the pathogenesis of FS or its recurrencesrelative to other fever-causing pathogens.

Knudsen (1988) showed that children experiencing four or more fever episodes peryear after a FS had a higher recurrence rate than children with up to three fever episodes ayear (78% vs. 17%), and child care in a nursery group compared to home day-care wasalso reported to increase the recurrence rate from 37% to 45% when the children werefollowed up for 18 months after the first FS (Knudsen 1985). Children under two years ofage have been shown to have a higher risk of respiratory infections in day-care centresthan if taken care of at home, and it is likely that the increase in the number of infectiouscontacts leads to more frequent feverish infections. The increase in the adjusted RR forrespiratory infections in day-care centres has been highest in one-year-old children, being1.69 (95%CI 1.43-2.01) for common colds, 1.99 (95%CI 1.57-2.52) for otitis media and9.69 (95%CI 3.31-40.55) for pneumonia (Louhiala et al. 1995). Half of all FSs areexperienced during the second year of life (Verity et al. 1985a), and thus day care at homeor in small groups might reduce the risk of their recurrence, but the hypothesis has notbeen tested.

2.3.3 Prostaglandins and cytokines in febrile seizures

Infective agents initiate an inflammatory reaction and PG synthesis in many tissues,including the central nervous system. The substrates for PG synthesis are phospholipids,which are converted to arachidonic acid by the enzyme phospholipase. Cyclo-oxygenaseenzymes (COX-1 and COX-2) convert arachidonic acid to a PG precursor, which isconverted to various PGs by specific PG synthetases (Vane et al. 1996). Severalendogenous PGs are produced in the brain (Seregi et al. 1985). PGs have a major role inelaborating fever. PGE2 at least has pyrogenic activity, but other arachidonic acidmetabolites may also contribute to body temperature regulation (Saper & Breder 1994).The levels of arachidonic acid and PGs are low in the mammalian brain under normalconditions (Wolfe & Mamer 1975), the highest basal concentrations of PGD2 andPGF2alfa in animal brains (although relatively low) having been found in thehypothalamus and cerebral cortex and lower concentrations in the cerebellum, medullaand hippocampal formation (Seregi et al. 1985).

It has been shown using MES or chemical induction of seizures in animal models thatthe concentrations of endogenous PGs increase in the brain under such conditions(Steinhauer & Hertting 1981, Förstermann et al. 1982, Navarro et al. 1989), the mostprominent rise being observed in PGD2 concentration, followed by PGF2alfa and otherPGs (Förstermann et al. 1982). The effect of PGs on seizures is controversial, sincedifferent PGs may have both provoking and inhibiting effects (Climax & Sewell 1981,Rosenkranz & Killam 1981). Tamai et al. (1983) compared the PGF2alfa levels in CSFafter a FS (31 cases) or an epileptic seizure (32 cases) or during a non-neurological

20

disease (20 cases) and found them to be higher in patients with a FS (mean 207.8, SE 50.7pg/ml vs. 43.0, SE 4.6 in those with an epileptic seizure and 53.3, SE 8.7 pg/ml in thosewith a non-neurological disease). Löscher and Siemes (1988) reported eight times higherPGE2 concentrations in CSF after a FS (17 cases) than for afebrile children (9 cases), butthe difference between the FS cases (mean 101, range 15-190 pg/ml) and 12 cases withfever and no seizures (mean 66, range 3-224 pg/ml) was not statistically significant.Linear regression analysis gave a significant positive correlation between PGE2concentration in the CSF and body temperature (p<0.001). The authors state that it is notclear whether the increased concentrations of PGE2 in the CSF after FS is due to theseizure or solely related to the febrile state of the child.

Cytokines, soluble proteins or glycoproteins that act as chemical communicatorsbetween cells and as inflammatory mediators, but not as effecting molecules, areproduced in leukocytes and other affected cells during infection or aseptic inflammation(Callard & Gearing 1994, Saper & Breder 1994). One specific cytokine often has severaltarget cells and biological activities, and the cytokines have been shown to have aprofound effect on leukocyte migration and function, haematopoietic cells, temperatureregulation, acute phase reaction, tissue remodelling and cell survival (Callard & Gearing1994). Interleukin–1beta (IL-1beta) has been shown to act as the predominantendogenous pyrogen in the fever reaction caused by pox-virus infection in mice (Alcami& Smith 1996), its effect evidently being linked to PG synthesis, as it is reduced by PGsynthetase inhibitors (Ushikubi et al. 1998). Although several other cytokines (IL-1alfa,IL-6, TNFalfa, interferons) have been implicated, their role in fever pathogenesis hasremained unclear (Saper & Breder 1994, Licinio & Wong 1996).

The level of IL-1beta in plasma is increased acutely after a FS relative to children withfever alone (Ichyama et al. 1998, Tutuncuoglu et al. 2001), and IL-1beta production bylipopolysaccharide (bacterial endotoxin)-stimulated blood monocytes in vitro is alsoenhanced after a FS relative to the situation in children with fever but no seizures(Helminen & Vesikari 1990). Concentrations of IL-1beta and other cytokines (IL-1alfa,IL-6 and TNFalfa) have not been found to be increased in the CSF of patients with FScompared with feverish children without seizures, but elevated concentrations have beenrelated to encephalitis and encephalopathy (Ichyama et al. 1998).

It is not clear how the effect of IL-1beta and other possible endogenous pyrogens ismediated from peripheral tissues to the central nervous system, where they stimulate PGsynthesis during inflammatory states. One theory suggests a transport system forcytokines through the blood-brain barrier (Licino & Wong 1996), while Saper and Breder(1994) state that they may reach the brain tissue through circumventricular organs, whichare specialized areas along the cerebral ventricular surface that have no blood brainbarrier. The circumventricular organ located near the thermoregulatory centre in thehypothalamus is the organum vasculosum, and this is thought to be important in feverpathogenesis. As a fever reaction develops rapidly after systemic administration of IL-1beta, it is also possible that a neuronal pathway may contribute to fever initiation.Nervus vagus stimulation by either IL-1beta or lipopolysaccharide, which is a bacterialendotoxin, increases production of the fever mediator PGE2 in the anterior hypothalamus,a reaction which has been shown in an animal model to be inhibited by vagotomy(Blatteis & Sehic 1997).

21

2.4 Prevention of recurrences

In the light of the benign nature of FS, the prevention of recurrences is aimed chiefly atalleviating the symptoms and reducing parental anxiety. Continuous anti-epilepticmedication with phenobarbital has been widely used since the first report of its efficacy inpreventing FS recurrences (Faero et al. 1972), in which children having a mean serumphenobarbital concentration of 16-30 mg/l during follow-up had a significantly lower rateof recurrences (4% vs. 21%) than those with a mean serum concentration of 8-15 mg/l.Correspondingly, Mamelle et al. (1984) found that one out of 22 patients receivingvalproate (4.5%), four out of 21 treated with phenobarbital (19.0%) and 9 of 26 in aplacebo group (34.6%) had recurrences. In the meta-analysis by Newton (1988) bothphenobarbital and valproate were ineffective in preventing recurrences, but only one outof the seven studies analysed had been randomized and placebo-controlled. Childrenassigned to daily phenobarbital after a FS have been shown to have a mean IQ 5.2 pointslower than placebo-treated children six months after discontinuation of the drug (Farwellet al. 1990), and as valproate can potentially cause severe fatal pancreatitis and hepatitis(Committee on Drugs 1982), neither can be recommended as preventive treatment forbenign FSs at present. Both carbamazepine and phenytoin have proved ineffective forpreventing recurrences (Bacon et al. 1981, Antony & Hawke 1983).

The findings on the effect of diazepam administered during fever on recurrences ofFSs have been controversial. Intermittent diazepam was found to be effective in oneplacebo-controlled study (Rosman et al. 1993) and in two trials in which it was comparedwith continuous phenobarbital or no therapy at all (Knudsen & Vestermark 1978,Knudsen 1985). One placebo-controlled study showed no effect of diazepam inpreventing FS recurrences (Autret et al. 1990).

High fever markedly increased the rate of recurrences (Berg et al. 1995, Rantala et al.1995). Antipyretic agents are routinely given to sick children who have a history of FS, toreduce the fever and hence (theoretically) the likelihood of recurrences (Baumann &Duffner 2000). Regular administration of acetaminophen (15-20 mg/kg x 6) has not beenfound to prevent early recurrences of FS, however, or to affect the occurrence or height offever as compared with sporadic administration (Schnaiderman et al. 1993).

2.5 Outcome of febrile seizures

The occurrence of epilepsy after a FS is 2-3%, which is higher than the figures of 0.4-1%reported among children with no FS (Nelson & Ellenberg 1976, Verity & Golding 1991,Berg & Shinnar 1996). The distribution of epilepsy by subtypes, including TLE, has beenfound to be similar in patients with FS and the rest of the population (Verity & Golding1991), and it has been stated that FSs are not likely to cause permanent brain damage orlead to unprovoked seizures. The increased prevalence of epilepsy after FS can beexplained by the ability of fever to trigger seizures in general. In a situation like this thefirst manifestation of epilepsy is recorded as FS.

22

Complex features in the initial FS have been reported to increase the risk ofrecurrences up to nine-fold, with each recurrence adding 65% to the risk of unprovokedseizures (Verity & Golding 1991, Berg & Shinnar 1996). Since these studies includedchildren with neurological problems, the confounding factors make interpretationdifficult. Previous neurological or developmental problems increased the risk ofunprovoked seizures to 35% (Berg & Shinnar 1996) and both complicated FSs andneurodevelopmental abnormality have been shown to be more common in patients withrecurrences (Verity et al. 1998).

FSs are said to have no adverse effects on the long-term intellectual and behaviouraloutcome, even for patients with a complex initial attack and recurrences (Ellenberg &Nelson 1978, Knudsen et al. 1996, Verity et al. 1998).

2.5.1 Mesial temporal sclerosis

Mesial temporal sclerosis (MTS) is a disease of the limbic system in which thehippocampal formation is the most profoundly affected structure, so that it is often termedhippocampal sclerosis, or Ammon’s horn sclerosis. Falconer et al. (1974) have performedan en-bloc dissection of the affected temporal lobe on hundreds of patients with drug-resistant epilepsy and found MTS in about half of the cases. In a series of 30 childrenwith TLE undergoing temporal lobectomy, 20 had MTS, of whom 15 had a history of FS,compared with one case among the 10 patients with no MTS (Falconer 1974). Numerousstudies have suggested that TLE patients with a history of FS have more frequent andmore intense sclerotic changes in mesial structures than TLE patients with no FSbackground (Kuks et al. 1993, Cendes et al. 1993, Harvey et al. 1995, Free et al. 1996,Barr et al. 1997, Marsh et al. 1997, Theodor et al. 1999), whereas only a few publishedreports have failed to find such a connection (Lawson et al. 1997, Salmenperä et al.1998). All the studies pointing to a connection between FSs and MTS have beenretrospective and their patients selected (severe cases of TLE).

VanLandingham et al. (1998) reported a high incidence of hippocampal intensity orvolume changes in MRI immediately after a complicated FS (in 10 out of 27 cases). Sincethe patients had a high incidence of abnormalities in their developmental status (22%), thepatients in this survey, too, may be regarded as selected.

There are no studies evaluating the occurrence of MTS in unselected FS patients, letalone ones in which the status of the temporal structures is reported, so that there is notrue evidence to support the speculated causal relation between FSs and MTS.

3 Aims of the research1. To study the literature on prostaglandins and seizures in order to clarify the possible

role of prostaglandins in febrile seizure pathogenesis.2. To evaluate risk factors for recurrence of febrile seizure and the prevention of such

recurrences. 3. To evaluate whether febrile seizures cause mesial temporal sclerosis.

4 Material and methods

4.1 Structure of the research

This thesis is based on five original publications (I-V). The role of PGs in thepathogenesis of FS is discussed in a systematic review of the effects of these and theirsynthetase inhibitors on seizures (I). To find new ways of reducing the risk of recurrencesof FS, we reviewed the literature on the medical prevention of recurrences (III) andanalysed and quantified the effects of those risk factors for recurrences that wereamenable to intervention (II) in a randomized, double-blinded, placebo-controlledprophylactic treatment trial with diazepam and acetaminophen (IV). The occurrence ofMTS in FS patients was studied by MRI-volumetry in an assessment of the long-termoutcome (V).

4.2 Methods for the review of prostaglandins and their synthetase inhibitors in relation to seizures

The aim of this systematic review of the role of PGs and their synthetase inhibitors withrespect to FSs was to examine all the experimental and human clinical studies dealingwith seizures and endogenous PGs or their synthetase inhibitors published in English. Weperformed a literature search using the Medline database using the keywords “seizure”,“convulsion” or “febrile seizure” and “prostaglandin”, “prostaglandin antagonist” or“prostaglandin inhibitor” to find the previously published data. All suitable articlesknown to us before the computer search were included, and the references of the newlyidentified ones were reviewed to ascertain that all relevant articles had been found. Allthe authors read the articles independently and analysed them using a predesigned datacollection sheet.

Altogether 68 articles on the role of endogenous PGs and PG inhibitors in seizureswere found and reviewed. Thirty-one of these were original works evaluating the effect ofPGs and/or their inhibitors on seizures, and were included in our systematic review.

25

Initially our aim was to make a meta-analysis of the available literature, but the diversityof the published articles and the lack of detailed original data made this impossible. Thusthe results were introduced in the form of a systematic review and tabulated withoutsummarizing them in the form of a synthesis.

4.3 Subjects and methods in the prophylactic treatment trial and the study of risk factors for recurrences

4.3.1 Subjects

All children from the primary catchment area of Oulu University Hospital are sent to theDepartment of Paediatrics after the first FS. The parents of those children treated at theDepartment of Paediatrics in that hospital for their first FS between November 1986 andSeptember 1990 were offered the opportunity of participating in a 2-year follow-uptreatment trial. The parents of all but three patients accepted, and only those patients whowere unable to attend the final 2-year follow-up visit subsequently had to be excluded.The eligibility criteria for the treatment trial were no previous seizures, no anticonvulsantmedication, no progressive neurological disease, no intracranial infection and no chronicdisease requiring continuous medication.

4.3.2 Treatment protocol

The calculation of the sample size needed for the treatment trial was based on previousfindings of a recurrence rate of 20% in our earlier febrile seizure surveys. A reduction to5% was regarded as clinically important. For a power of 80% and a type I error of 0.05,the calculated sample size for each group was 66 children. To ensure eventual samples ofthis size, we included 90 patients in each anticonvulsive treatment group. Thus the totalnumber of children needed for the treatment trial was 180.

The participants were randomly assigned to receive either a placebo or diazepamduring all their febrile infections, employing blocks of varying sizes of four, six, or eightpatients within each stratum. Only the biostatistician knew the details of therandomization schedule.

The medication was started with a single dose of diazepam or placebo solution forrectal use (rectiol) when the body temperature exceeded 38.5°C. The dosage of diazepamwas 2.5 mg for children weighing less than 7 kg, 5 mg for children weighing from 7 to 15kg and 10 mg for children weighing more than 15 kg. Six hours after receiving the rectiolthe medication was continued with oral diazepam doses of 0.2 mg/kg or placebo solutionthree times a day for the first two days if the child was still feverish (Fig. 1).

26

Fig. 1. Randomization of patients into groups for prophylactic anticonvulsive medication(diazepam and placebo) during all subsequent fever episodes and for antipyretic treatment(acetaminophen and placebo) during the first fever episode. The assignment to acetaminophenor placebo was switched after each fever episode.

The first febrile episode was assigned randomly to management with an oral solutionof acetaminophen or placebo, the assigned treatment then being switched after eachfebrile episode (Fig 1). The dose of acetaminophen was 10 mg/kg four times per day.

The dosage was adjusted according to changes in the body weight of the child duringthe 2-year follow-up. If the child’s temperature exceeded 40°C the parents were allowedto give extra acetaminophen 10 mg/kg. If a FS occurred, it was treated with a diazepamrectiol and the medication was stopped.

The study was double-blinded. The final analysis included febrile events in foursubgroups defined according to the assigned treatment, i.e. diazepam and acetaminophen,diazepam and placebo, placebo and acetaminophen and two placebos (Fig. 1).

4.3.3 Evaluation and follow-up of the patients

The participants had been examined at the time of the hospitalization for the first FS bythe duty physician, who had made a basic evaluation of their neurological anddevelopmental status and recorded the symptoms and signs of infection. The parents wereinterviewed for background information.

The parents of the 180 patients included in the treatment trial filled in a questionnairewhenever their child was feverish during the 2-year follow-up or if they suspected anunprovoked seizure. The information of interest was the type and timing of thesymptoms, the rectal temperature, recorded twice a day during the febrile illness, and theuse of the study medication and/or other antipyretics. A nurse contacted the participatingfamilies monthly in order to receive all the information on possible infections and tomotivate the parents to follow the protocol.

Anticonvulsive Placebo as an anticonvulsive drug for each episode

Group 1 Antipyretic Placebo Acetaminophen Placebo

Eliglible patients 1 2 3 etc… randomized Anticonvulsive Diazepam as an anticonvulsive drug for each episode Group 2

Antipyretic Acetaminophen Placebo Acetaminophen

1 2 3 etc…

Number of febrile episodes

27

A thorough neurological and developmental evaluation was performed by a paediatricneurologist at the end of the 2-year follow-up, including basic developmental status(height, weight, head circumference), motor and sensory function tests, visus and motorfunction of the eyes, speech and hearing. An EEG was also taken.

Of the total of 180 patients, 161 were followed up for the 2-year period, 157 came tothe last outpatient examination and an EEG was obtained for 156. The final analysis ofthe efficacy of the drugs was based on data for 153 children who had had at least onefebrile episode and had thus used the medication supplied. Additionally, all the availabledata on the 10 patients who dropped out after 1 to 18 months of the follow-up who hadhad at least one prospectively followed febrile episode were included in the Kaplan-Meieranalysis.

4.3.4 Risk factors for recurrences

Baseline information on the 180 children participating in the treatment trial and their finalphysical, developmental and neurological assessments, the background information fromthe parental interviews and the prospective follow-up data on each febrile episode duringthe two-year period were used in the analysis of the risk of recurrence and the short-termoutcome.

The 65 girls and 88 boys who had had at least one febrile episode during the follow-upperiod, at a mean age (range) of 1.7 (0.8 to 4.6) years, were included in the risk factoranalysis. The influence of age at enrolment on the rate of FS recurrence was analysedusing the age at the time of the initial seizure as a continuous variable. The initial seizurewas a complicated one in 49 children. A positive family history of FS was found in 59children, and 18 had a positive family history of epilepsy. The influence of temperatureduring the subsequent fever episodes on the recurrence of FS was analysed by calculatingthe mean of the maximum temperatures during these episodes in the children withoutfurther seizures and comparing them with the means of the maximum temperatures of thefever episodes associated with a FS in the children with one or more recurrences.

The data on the 156 children for whom an EEG was obtained at the 2-year follow-upvisit were included in the analysis of the effect of recurrences on the short-term outcome.

4.4 Meta-analysis of prophylactic treatment against recurrences

To find all existing randomized placebo-controlled studies of the prevention of FSspublished in English, we performed a literature search using the Medline database. Thepreventive treatment trials known to us beforehand were also accepted, and to ensure thatall the relevant articles had been found, the references in the accepted papers werereviewed. Altogether we found and reviewed 45 articles for the meta-analysis. All theauthors read the articles independently and analysed them using a predesigned datacollection sheet. Nine out of the 45 articles filled the criteria for a randomized, placebo-controlled treatment trial.

28

4.5 Patients and methods in the mesial temporal sclerosis survey

The connection between FSs and MTS was studied in a sample from the 329 unselectedFS patients who had participated in our clinical evaluation of the factors triggering thefirst FS, the risk factors for recurrences and the prevention of recurrences at theDepartment of Paediatrics, University of Oulu, during the years 1984 to 1990 (Rantala etal. 1990, Rantala et al. 1994, Uhari et al. 1995). The opportunity to participate in theoutcome study, including MRI of mesial temporal structures and a neurologicalevaluation, was offered to the 30 patients with a prolonged initial FS and the eightpatients with at least one unprovoked seizure after the first FS. One patient who met bothcriteria was analysed in the unprovoked seizure group. All the patients with anunprovoked seizure participated, but three patients in the prolonged FS group could notbe reached and three others chose not to participate. For each of the 32 cases we selectedan age, sex and handedness-matched control patient among those who had had a singlesimple FS with no recurrences or unprovoked seizures. Out of the eight patients in theunprovoked FS group, three had had complex partial seizures, two had rolandic epilepsy,one had myoclonic seizures, one had had several focal secondarily generalized seizuresand one had experienced a single unprovoked seizure with secondary generalisation. Themean age (range) of the patients with a prolonged initial FS at the time of the MRIexamination was 14.4 (9.9-20.2) years, that of the patients with later unprovoked seizures12.5 (10.4-14.2) years and that of the controls 14.2 (10.3-20.4) years. The mean follow-up times (range) in these groups were 12.5 (8.5-14.7) years, 11.2 (8.9-12.6) years and12.5 (9.6-14.7) years, respectively.

The patients or their parents were asked about previous seizures and medical history,scholastic achievements and problems in learning. The hospital records of the participantswere reviewed, and a clinical examination was performed, including developmentalstatus, i.e. height, weight, head circumference and Tanner pubertal stage (Tanner &Whitehouse 1976), motor and sensory function tests, visus and motor function of theeyes, speech and hearing.

MRI was performed using a 1.5 Tesla scanner (Signa, EchoSpeed, General ElectricMedical Systems, Milwaukee, Wis), obtaining T1-weighted sagittal images together withdouble fast spin echo T2-weighted axial and coronal slices. The T2-weighted axialimages were obtained parallel to the temporal lobes and the coronal images perpendicularto them. A 3D coronal SPGR series was also obtained, providing high grey matter andwhite matter contrast, and transferred to a workstation for volumetry. Reformatted imagestwo millimetres thick were generated perpendicular to the hippocampal formations, andthe volumes of both the amygdala and the hippocampal formations were measured onthese images by one radiologist who was blinded to the clinical history of the subjects.The boundaries of the structures concerned were defined according to previous reports(Watson et al. 1992). The in-house software used for this employs a semi-automatedtechnique combining tracing and a threshold. All the MR images were also evaluatedvisually by two radiologists, first separately and then together, to reach a consensus.Special attention was paid to the size, shape and signal intensity of the hippocampalformations.

Since there are no normal values for adolescent patients, we used the findings in ourcontrol group, i.e. the patients with a single simple FS, as a source for reference values.

29

The right-left hippocampal volume difference threshold values in our adolescent controlpatient group (2 SD range) were –0.13 cm3 and 0.49 cm3.

4.6 Ethical considerations

The protocols were approved by the ethical committee of the Medical Faculty of theUniversity of Oulu, and informed consent was obtained from a parent or the patient at thebeginning of each phase of the research.

4.7 Statistical methods

SPSSWin versions 6 and 9 were used to analyse the data.The relative risk for each recurrence factor was calculated separately and a

multivariate risk factor analysis was performed using the Cox proportional-hazard model.The effect of recurrences on the outcome was analysed by logistic regression.

In the treatment trial the frequency of recurrences of FS in the patients receivingdiazepam was compared with that in the placebo group. The effect of acetaminophen wasevaluated during each fever episode separately in numerical order by comparing thefrequency of recurrences in patients receiving acetaminophen with those in patientsreceiving the placebo during the episode in question, and the frequency of recurrencesduring all the episodes treated with acetaminophen was also compared with that in theepisodes with placebo treatment. The differences in the frequencies of recurrences of FSbetween the patient groups and between the fever episodes were analysed with the chi-square test. The times to first recurrence in these strata were assessed by the Kaplan –Meier method. The cumulative incidence curves for recurrences during the two-yearfollow-up were compared between the groups with the log-rank test.

The mean absolute volumes of the hippocampal formation and amygdala and the meanright-left volume difference in the hippocampal formations were compared between thegroups using a one-way analysis of variance. In the case of a statistically significantdifference, the analyses were continued using Bonferroni Post Hoc multiple comparison.The Kruskal-Wallis test was used to compare the number of patients with problems inschool performance or abnormal neurological findings between the groups. Thesignificance of a difference between two parametric variables was assessed by the t-test.

The meta-analysis of preventive treatment was performed by combining the originaldata for each treatment modality considered. A test of heterogeneity was performed onthe total set of papers before combination of the data (Thompson 1994). The childrenwith recurrences were regarded as treatment failures. The weighted average risk wascalculated using the inverse of the variance attached to the risk in each paper as theweight (random effect model) (Hedges 1994). 95% confidence intervals were calculatedfor the odds ratios, and the Mantel-Hazel chi-square test was used to assess theirstatistical significance. The effect of each treatment modality was reported in terms of thenumber of patients needed to treat to prevent a FS recurrence, and 95% confidenceintervals were calculated where the efficacy of treatment was statistically significant.

5 Results

5.1 Endogenous prostaglandins, prostaglandin synthetase inhibitors and seizures

The original six articles on the seizure modulating effects of various PGs in humans werereports on extra-amniotic or intra-amniotic PGE2 or PGF2alfa administered to induceabortion (Lyneham et al. 1973, MacKenzie et al. 1973, Fraser & Gray 1974, Shearman etal. 1975, Faden et al. 1976) or intravenous or oral PGE2 or PGF2alfa for the induction oflabour (Thiery et al. 1974). All five major seizures reported in relation to PGadministration occurred in the series described by Lyneham et al. (1973), in which 320women had received intra-amniotic PGF2α for the induction of abortion.

There were 13 articles evaluating the effect of endogenous PGs on chemically orelectrically induced seizures, and 12 that of PG synthetase inhibitors. All the experimentshad been conducted with adult animals.

Intracerebroventricular or intraperitoneal administration of PGD2, PGE1 or PGE2prior to seizure induction increased the latency of the onset of seizure and/or reduced theseizure incidence, and thus had an anticonvulsant effect (Rosenkranz & Killiam 1979 and1981, Climax & Sewell 1981, Förstermann et al. 1983). These PGs did not have anyeffect on seizures caused by kindling (Croucher et al. 1991). The effect ofintracerebroventricular PGF2alfa was controversial, being said in one paper to haveinhibited pentylenetetrazole (PTZ)-induced seizures (Förstermann et al. 1983) and in twoothers to have promoted seizures induced either by PTZ or maximal electroshock (Climax& Sewell 1981, Rosenkranz & Killam 1981) (Table 1).

Twelve evaluations of the effect of non-steroidal anti-inflammatory drugs (NSAIDs),i.e. PG synthetase inhibitors, on seizures in experimental animals have shown the mostpowerful inhibitors of cyclo-oxygenase (COX) enzymes, diclofenac, flurbiprophen andindomethacin, to either promote chemically or electrically induced seizures or to have noeffect on them when administered by either an intramuscular or intraperitoneal route(Steinhauer & Hertting 1981, Förstermann et al. 1982, Wallenstein & Mauss 1984). It isinteresting that acetylsalicylate, which is the least potent inhibitor of COX enzymesamong the NSAIDs tested, promoted seizures induced with both maximal electroshock

31

and PTZ in one trial (Climax & Sewell 1981). Other less potent COX inhibitors,ibuprofen, mefenamic acid, meclofenamic acid, acetaminophen and sulindac, had mainlyinhibitory effects on seizures induced with PTZ, penicillin and flurothyl (Wallenstein &Mauss 1984, Wallenstein 1987), but no effect on those induced by maximal electroshock(Wallenstein & Mauss 1984) (Table 2).

Table 1. Effects of certain prostaglandins (PG) on chemically or electrically inducedseizures in adult animals.

PG Route of administra-tion

Dosage(µg)

Seizureinduction

Animal Number of animals with PG

Controls Effect on seizure

Paper

D2 Amygdala 1–10 Kindling Rat 9 15 None 1ICV 2–20 PTZ Rat 71 124 Inhibitory 2, 3, 4

E1 ICV, IP 0.5–60 PTZ Rat, mouse 96 118 Inhibitory 5, 6, 7, 8ICV, IP 1–320 MES Rat, mouse 130 110 Inhibitory 8, 9, 10IP 0.6–50 STR Mouse 104 104 Inhibitory 7, 11ICV 0.5–20 PTX Mouse 70 70 Inhibitory 7ICV 1–30 INZ Mouse 60 60 Inhibitory 7

E2 Amygdala 1–10 Kindling Rat 15 15 None 1ICV 1–160 PTZ Rat, mouse 168 203 Inhibitory 2, 7, 12, 13ICV, IP 2–320 MES Rat, mouse 230 210 Inhibitory 9, 10, 13ICV 1–40 STR Mouse 140 140 Inhibitory 10, 13ICV 1–50 PTX Mouse 140 140 Inhibitory 10, 13ICV 2–50 INZ Mouse 120 120 Inhibitory 10, 13ICV 12.5–100 FLS Baboon 6 6 Inhibitory 13

F2alfa Amygdala 1–10 Kindling Rat 9 15 None 1ICV 20 PTZ Rat 20 47 Inhibitory 2ICV 10–60 PTZ Mouse 10 10 Promoting 8ICV 10–60 MES Mouse 100 60 Promoting 8, 13

FLS, flashing light stimulation; ICV, intracerebroventricular; INZ, isoniazid; IP, intraperitoneal; MES, maximalelectroshock; PTZ, pentylenetetrazole; PTX, picrotoxin; STR, strychnine; 1, Croucher et al. 1991; 2, Förster-mann et al. 1983; 3, Bhattacharya & Parmar 1987; 4, Akarasu et al. 1998; 5, Bodzenta & Wisniewski 1977; 6,Bhattacharya & Sanyal 1978; 7, Rosenkranz & Killam 1979; 8, Climax & Sewell 1981; 9, Madan et al. 1974;10, Rosenkranz 1978; 11, Duru & Turker 1969; 12, Berti et al. 1976; 13, Rosenkranz & Killam 1981

32

Table 2. Effects of prostaglandin synthetase inhibitors (PGSI) on experimentally inducedseizures in adult animals.

PGSI Route of PGSI admin-istration

Dosage Seizure induction

Animal No. of animals with PGSI

Con-trols

Effect on seizure

Paper

Acetylsali-cylate

Oral 100 mg/kg PTZ Mouse 8 10 None 1

ICV 5–40 µg MES, MES Mouse 80 20 Promoting 2Diclofenac IM 0.1–33 mg/kg PTZ Mouse 400 200 None 3

IM 10 mg/kg PTZ Mouse 9 10 Promoting 1Flurbiprofen IM 0.1–10 mg/kg PTZ Mouse 150 150 None 3

IM 10 mg/kg PTZ Mouse 24 10 Promoting 1Ibuprofen IM, IP 10 mg/kg PTZ Rat, mouse 34 35 None 1, 4, 5

IP 10–90 mg/kg MES, BI, PTX

Rat 54 24 None 5

IP 10–90 mg/kg FL, PTZ Rat 44 26 Inhibitory 4, 5, 6, 7Indomethacin IP 2–30 mg/kg MES, FL,

PTX, PTZRat 52 52 None 5, 6

IM, IP 0.1–100 mg/kg

PTZ, stress Mouse,gerbil, rat

547 484 Promoting 1, 3, 8–12

Mefenamic acid

IP 15–50 mg/kg MES, BI, FL, PTX

Rat 46 36 None 4, 5

IP 20–150 mg/kg FL, PN, PTZ

Rat 16 16 Inhibitory 5, 7

IP 60 mg/kg PTZ Rat 6 – Promoting 4Meclofenamic acid

IP 15–50 mg/kg MES, BI, FL, PTX

Rat 54 31 None 5

IP 15–50 mg/kg PTZ Rat 8 8 Inhibitory 5, 6Acetami-nophen

IM, IP 30–300 mg/kg MES, BI, FL, PTZ

Rat, mouse 70 77 None 1, 4, 5

IP 150–450 mg/kg

PN, PTZ Rat NA NA Inhibitory 6, 7

Sulindac IP 15–150 mg/kg MES, FL, PTX

Rat 49 40 None 5

IP 50–150 mg/kg FL, PTZ Rat 24 20 Inhibitory 5BI, bicuculline; FL, flurothyl; ICV, intracerebroventricular; IM, intramuscular; IP, intraperitoneal; MES, maxi-mal electroshock; PN, penicillin; PTZ, pentylenetetrazole; PTX, picrotoxin; NA, not available; 1, Steinhauer &Hertting 1981; 2, Climax & Sewell 1981; 3, Förstermann et al. 1982; 4, Wallenstein 1991; 5, Wallenstein &Mauss 1984; 6, Wallenstein 1985; 7, Wallenstein 1987; 8, Steinhauer et al. 1979; 9, Förstermann et al. 1984; 10,McGinley et al. 1985; 11, Seregi et al. 1984; 12, Simmet & Tippler 1990

33

5.2 Risk factors, prognosis regarding recurrences and results of the treatment trial

5.2.1 Occurrence of fever episodes and seizures during follow-up

There were altogether 642 febrile episodes and 55 FS recurrences during the two-yearfollow-up. Thirty-eight patients had at least one recurrence and 11 experienced multiplerecurrences. Fifteen had a FS recurrence during their first febrile event, and 32 had arecurrence within one year of the initial FS. Two patients experienced an unprovokedseizure during the follow-up.

5.2.2 Risk factors and the prognosis regarding recurrences

The multivariate risk factor analysis showed the number of febrile episodes and thedegree of fever during the subsequent febrile episodes to have a statistically significantinfluence on the recurrence rate. Each febrile episode increased the risk of recurrence by18% (RR 1.2, 95% CI 1.1-1.3, p = 0.0003) and each degree of increase in bodytemperature during the subsequent fever episodes almost doubled the recurrence risk (RR1.9, 95% CI 1.0-3.7, p = 0.04). A high temperature during the initial FS did notsignificantly reduce the risk of recurrences, but it had a low risk ratio (RR 0.7, 95% CI0.4-1.1, p = 0.13). A family history of FS, a family history of epilepsy, the gender and ageof the child and the type of initial FS were included in the risk factor analysis but theirinfluence on the recurrence rate was insignificant. The univariate risk factor analysis andthe multivariate Cox proportional Hazard model gave similar results.

Assessment of the effect of recurrences on the short-term outcome showed six of the156 patients to have developed defects in speech, seven abnormal findings in motorfunction tests and seven disturbances in eye movements and vision. Sixteen children hadeither spikes or spikes and waves in their EEG recording. No statistically significanteffect of recurrences on these variables was found.

5.2.3 Effect of acetaminophen and low dose diazepam on recurrences

The prophylactic treatment trial showed acetaminophen, low intermittent doses ofdiazepam and their combination to be ineffective for preventing recurrences of FS ascompared with a placebo.

Of the 79 children assigned to receive acetaminophen during the next febrile episode,eight experienced a FS recurrence during that episode, whereas seven recurrences wererecorded among the 74 children in the placebo group (Table 3).

34

Table 3. Numbers of patients and recurrences of FS during the next fever episode after theinitial FS, by mode of treatment.

There were altogether 23 FS recurrences during 314 acetaminophen-treated feverepisodes during the 2-year follow-up (7.5%) and 32 recurrences during 327 placebo-treated fever episodes (9.8%) (Table 4). Thus acetaminophen had no effect on recurrencesof FS.

Table 4. Numbers of fever episodes and recurrences during the 2-year follow-up by modeof treatment.

Parents gave similar amounts of extra antipyretic agents to the children receivingacetaminophen as to those receiving placebo treatment. The mean temperature duringeach febrile episode was 0.5°C higher in those children who had seizures than amongthose who did not, but the mean difference in the highest temperatures for all febrileepisodes between the acetaminophen and placebo groups was not significant (difference0.3°C, p >0.05). Even though acetaminophen reduced the mean temperatures, it had noeffect on the highest temperatures or on the temperatures in those patients who had FSs.

Of the 153 patients experiencing febrile events during the 2-year follow-up, 74 hadbeen assigned to receive diazepam and 79 to receive placebo treatment during allsubsequent fever episodes. The numbers of patients with at least one recurrence duringfollow-up in these groups were 21 (28.4%) and 17 (21.5%), respectively. No statisticallysignificant difference in the recurrence rates could be found between these groups, nor didthe Kaplan-Meier method show any difference in the timing of the first recurrence (Fig2).

Treatment duringnext fever episode

No. of patientsin treatment group

No. of recurrences (%) during next fever episode

Diazepam + Placebo 35 4 (11.4)Placebo + Acetaminophen 40 2 (5.0)Diazepam + Acetaminophen 39 6 (15.4)Placebo + Placebo 39 3 (7.7)All treatment groups 153 15 (9.8)The differences in recurrence rates during the next fever episode between the children assigned to diazepam,acetaminophen or both and those receiving only placebo treatment were not statistically significant.

Treatment during fever episodes No. of fever episodes No. of recurrences (%) during fever episodes

Diazepam + Placebo 156 18 (11.5)Placebo + Acetaminophen 173 9 (5.2)Diazepam + Acetaminophen 141 14 (9.9)Placebo + Placebo 171 14 (8.2)All treatment groups 641 55 (8.6)The differences in recurrence rates during the fever episodes between the patients assigned to diazepam, aceta-minophen or both and those assigned to placebo treatment were not statistically significant.

35

Fig. 2. Cumulative incidence of recurrent FS, by anticonvulsive treatment group(–––Diazepam, ---- Placebo). The difference between the curves was not significant (P=0.4138).

Similarly, no difference in recurrence rates was observed between the diazepam andplacebo groups during the next fever episode (Table 3) or any subsequent episode(considered in rank order).

The combination of antipyretic and anticonvulsive medication did not reduce therecurrences of FS compared with any other treatment regimen used (Table 4).

5.3 Meta-analysis of preventive treatment for febrile seizure recurrences

Of the nine articles accepted in the meta-analysis, four evaluated the effect of continuousphenobarbital on recurrences of FS in a placebo-controlled setting (Camfield et al. 1980,Bacon et al. 1981, Mamelle et al. 1984, Farwell et al. 1990), one of them also evaluatingthe prophylactic effect of valproate (Mamelle et al. 1984) and another phenytoin (Baconet al. 1981). One paper compared the effect of intermittent phenobarbital with placebotreatment (Mackintosh 1970), three evaluated the prophylactic effect of diazepam (Autretet al. 1990, Rosman et al. 1993, IV) and one considered the effect of pyridoxine onrecurrences (McKiernan et al. 1981). The test of heterogeneity applied to the dataconcerning each treatment showed this to be insignificant, justifying the combination ofthe material.

The meta-analysis revealed a significant reduction in the FS recurrence rate in childrenreceiving continuous phenobarbital relative to placebo treatment (OR 0.54, 95% CI 0.33-0.90, p = 0.017), whereas the only assessment of intermittent phenobarbital treatment

242220181614121086420

0,4

0,3

0,2

0,1

00,0

CumulativeIncidence

36

relative to a placebo showed no difference in the risk of recurrence between the groups(OR 0.51, 95% CI 0.10-2.62, p = 0.41). The combined results of the three trials evaluatingthe effect of diazepam administered during fever episodes showed no effect of thisintermittent treatment on the FS recurrence rate relative to a placebo (OR 0.81, 95% CI0.54-1.22, p = 0.31). The one comparison of continuous valproate with a placebo showeda preventive effect on further recurrences (OR 0.09, 95% CI 0.01-0.78, p = 0.011). Bothpyridoxine and phenytoin were ineffective when compared with a placebo.

The number of patients that it was necessary to treat in order to avoid one recurrenceof FS was eight for phenobarbital and four for valproate. For the drugs with insignificantefficacy in the meta-analysis, 26 patients would have had to be treated with intermittentdiazepam, 119 with phenytoin, and 16 with pyridoxine to prevent one recurrence.

5.4 Results of the mesial temporal sclerosis survey

The results of the evaluation of the causal relation between FS and MTS were negative.The qualitative examinations of the mesial temporal area by MRI revealed no cases withsclerosis, and there were no statistically significant differences in absolute right or lefthippocampal formation or amygdala volumes between either the patients with aprolonged initial FS or those with an unprovoked seizure after the initial FS and thosewith a single simple FS. The mean right-left hippocampal volume difference wassignificantly smaller in the patients with a prolonged initial FS, i.e. 0.039 cm3 than in thecontrols, 0.18 cm3 (p <0.01) (Table 5), and the eight cases with a focal prolonged initialFS had an even lower mean difference, -0.081 cm3 vs. 0.18 cm3 (p <0.001).

Table 5. Results of MRI volumetry. Mean (SD) structural volumes, in cm3, for the childrenwith initial prolonged or simple FS and those with later unprovoked seizures.

Five children in the prolonged FS group had an initial attack lasting 60 minutes andone child 150 minutes, but their mean volume and volume difference values did not differfrom those of the patients with a prolonged seizure lasting from 30 to 59 minutes. Thelongest duration of a focal prolonged initial seizure was 60 minutes, observed in onechild. The mean hippocampal volume difference in the patients with an unprovokedseizure was similar to that for the controls. The results of the comparison between the

Variable Unprovokedseizure after

FS (n=8)

Prolonged initial

FS (n=24)

Single simple FS (=32)

P-value of differences between the groups

Right HF 2.95 (0.27) 2.83 (0.42) 2.98 (0.40) 0.34Left HF 2.83 (0.30) 2.79 (0.42) 2.80 (0.36) 0.96Right-left difference in HFs 0.12 (0.16) 0.039 (0.19) 0.18 (0.16) 0.012*Right amygdala 2.03 (0.28) 2.04 (0.24) 2.06 (0.30) 0.97Left amygdala 1.99 (0.26) 2.01 (0.26) 1.95 (0.28) 0.76HF, hippocampal formation, *Significant differences between patients with prolonged initial FS and the control group in Bonferroni multiple comparison: mean difference = -0.1411, 95% CI 0.254--0.028, p-value <0.01.

37

mean greater-smaller hippocampal volume differences for the groups were similar tothose of the analysis in terms of the mean right-left hippocampal volume difference.

According to our threshold values for the right-left hippocampal volume difference(relative to the values for our controls), three patients in the prolonged FS group had asmaller right hippocampal formation (right-left volume difference ranging from -0.47 cm3

to -0.15 cm3), while one left-handed control had a volume difference of 0.52 cm3. Allthese four patients were neurologically normal, without unprovoked seizures.

The school performance and neurological findings of the patients with a prolongedinitial FS were as good as those of the control group. The children with problems inschool performance or abnormal neurological findings had similar hippocampalformation and amygdala volumes and right-left volume difference in hippocampalformations to the patients with normal neurological status and no problems in schoolperformance. Three patients with a prolonged initial FS, three with a later unprovokedseizure and eight controls had mild abnormalities in the motor function tests, and threepatients in the prolonged FS group, two in the unprovoked seizure group and one controlhad a history of problems in school performance. There were no statistically significantdifferences in school performance between the patients with a prolonged primary FS,those with a later unprovoked seizure and the controls.

The mean absolute hippocampal and amygdala volumes and the right-left volumedifferences in the patients with multiple first seizures (three cases) and those with up toten recurrences during the two-year follow-up (four cases) were similar to the values forthe control patients.

6 Discussion

6.1 Study design

This work was originally started with a randomized, placebo-controlled trial evaluatingthe effect of diazepam and acetaminophen on recurrences of FS (IV). As this regimenlacked any effect in our trial (IV), a need arose for a further analysis of risk factors forrecurrences, especially factors that were amenable to intervention (II). Since at that timeno consensus had been reached on the prevention of recurrences, a meta-analysis of theliterature on medical prophylaxis against recurrences was conducted (III). The possiblerole of PGs in FS pathogenesis needs further consideration, as PG synthetase inhibitorsare widely used as forms of antipyretic treatment for children. To start with, weperformed a systematic review on this subject (I). Although FS are considered benign, acausal relation with MTS has been suggested. The FS patients that we used for theoutcome analysis with respect to MTS (V), was a representative sample of cases followedup prospectively for more than a decade.

6.2 Prostaglandins and their synthetase inhibitors in relationto febrile seizures

Our systematic review suggested that some endogenous PGs produced in the brain(PGD2, PGE1, PGE2) provide protection from seizures in animals, while others (notablyPGF2alfa) have mainly aggravating effects (Table1). PGF2alfa and PGE2 have been usedfor the induction of abortion and labour in women, but seizures have been reported inonly one series, where five out of 320 patients (including two with a history of epilepsy)had seizures after intra-amniotic administration of PGF2alfa (Lyneham et al. 1973). Acausal relationship between PGF2alfa or PGE2 infusion and seizures in normalindividuals is therefore unlikely.

Non-steroidal anti-inflammatory agents (NSAIDs) inhibit the multi-enzyme complexCOX and thus reduce PG synthesis. In our review diclofenac, fluriprofen andindometacin, which are strong inhibitors of the COX enzymes, mainly increased seizure

39

susceptibility, while ibuprofen, mefenamic acid, meclofenamic acid, acetaminophen andsulindac, possessing a weaker COX inhibition capacity, had an inhibitory effect on mostseizures induced chemically (with PTZ, flurothyl or penicillin), but no effect on thoseinduced electrically. Acetylsalicylate was an exception to this, provoking seizures eventhough it is a weak COX inhibitor (Climax & Sewell 1981). The anticonvulsive effect ofthe weak COX inhibitors (except acetylsalicylate) increased in a dose–dependent manner,but the strong COX inhibitors all provoked seizures. Thus the intensity of COX inhibitiondoes not fully explain the seizure susceptibility modulating effect of NSAIDs.

The PG synthetase inhibitors vary in structure and differ not only in the intensity oftheir inhibition of COX enzymes but also in their COX-1 and COX-2 enzyme inhibitionratio (Vane et al. 1996), which may explain their effect on seizure susceptibility. The roleof COX-2 activity in kainic acid-induced seizures in mice was evaluated by Baik et al.(1999), who found that selective inhibitors of this activity attenuated PG production andincreased neuronal death in the hippocampal formation most profoundly. The aggravatingeffect of unselective PG synthetase inhibitors on seizures was related to their COX-2inhibition intensity. These findings indicate a need for further studies of the possibleseizure inhibiting and neuroprotective effects of COX-2 activity and also offer a possibleexplanation for the effects of different COX inhibitors in relation to their COX-1 andCOX-2 inhibition intensity.

Elevated levels of PGF2alfa and PGE2 are found in the CSF of children after a FS(Tamai et al. 1983, Löscher & Siemes 1988). NSAIDs inhibit PG synthesis and areeffective antipyretic drugs, but their effect on seizures is not clear. If NSAIDs reducefever by inhibiting the synthesis of anticonvulsive PGs, the net effect on the risk of FS isunpredictable. Acetaminophen in our treatment trial (IV) and ibuprofen in that of vanStuijvenberg et al. (1998) were ineffective in preventing FSs, and similarly neitheracetaminophen nor ibuprofen prevented PTZ induced seizures at low doses in animalmodels and electrically (MES), bicuculline or flurothyl-induced seizures at any dose(Table 2). These findings imply that the pathogenesis of FS in children may bearsimilarities to these artificially induced seizures in animals. It is notable, however, thatthere are no studies on immature animals in this field, as all the experimental work in oursystematic review had been performed with adult animals. On the other hand, FSs onlyoccur in children, and epileptogenesis differs between the adult and immature brain. Sinceit is possible that PGs and their synthetase inhibitors act differently in immature animals,no direct conclusions about the effects of PGs and their synthetase inhibitors on seizuresusceptibility in humans can be drawn from these publications.

The diversity of methods used for the induction of seizures in experimental animalmodels in the review also made the comparison of the results and interpretation of thedata difficult. As FSs are usually generalized tonic-clonic seizures and MES-inducedseizures are considered a prototype model for these (Fisher 1993), MES-induced seizuresoffer the most suitable model for FS among those mentioned in the review. MES has notbeen used in experimental FS studies, where seizures have been induced by variousapplications of external heat (Millichap 1959, Chen et al. 1999), but these models do notoptimally mimic FS in humans.

40

6.3 Opportunities to modify risk factors for recurrences

The results of our risk factor survey corroborate previous findings in which frequent feverepisodes after the initial FS markedly increased the risk of recurrences (Knudsen 1988,Rantala & Uhari 1994). In our material, each fever episode after the initial seizureincreased the recurrence risk by 18% (p = 0.0003). The quantitative effect of febrileepisodes on recurrences has not previously been evaluated. High fever is the strongestfactor triggering the first FS (Rantala et al. 1995, Berg et al. 1995), and each degree ofrise in body temperature during subsequent fever episodes almost doubled the risk ofrecurrence in the current trial (RR 1.9, p = 0.043).

Young age at the onset of FSs was found to increase the recurrence risk in 12 out of the14 reports evaluated by Berg et al. (1992) in their meta-analysis. We did not find,however, that age reached statistical significance as a predictor of recurrences, perhapspartly because of the collinearity between age and the number of subsequent feverepisodes. The height of the temperature during the initial FS had a low risk ratio (RR 0.7)in our data, but its effect on the frequency of recurrences was not statistically significant(p = 0.13). The lack of a decrease in the risk of recurrences in association with hightemperature may well be due to the small sample size in our material. Several otherstudies have found a significant reduction in recurrences in children with high feverduring initial FS (Berg et al. 1992, El-Radhi & Banajeh 1989, Offringa et al. 1992).

Efforts to prevent febrile infections by restricting the number of infectious contactsmight theoretically reduce the risk of recurrence. The most effective way to restrict thenumber of infections in children would be to avoid large day-care centres or nurserygroups. Interventions to prevent the spread of infective agents in day-care centres havealso been shown to reduce infections (Uhari & Möttönen 1999). Whether child-care athome could prevent children with a history of FS from suffering further recurrences betterthan outside day-care is not known. Nevertheless, the fear of multiple or complicatedrecurrences, and parental anxiety in general, could lead to interventions in the type ofday-care.

6.4 Antipyretic drugs for preventing recurrences

As high fever is the major triggering factor for FS (Rantala et al. 1995, Berg et al. 1995)and antipyretic drugs acting via the PG system are widely used in children, the effect ofantipyretics on recurrences of FS should be evaluated. Although high amounts of PGsoccur in the CSF of children after FS relative to children with unprovoked seizures(Tamai et al. 1983), ibuprofen and acetaminophen seem not to prevent recurrences (vanStuijvenberg et al. 1998, IV). The low dose of acetaminophen used in our trial (40 mg/kg/24h) does not explain the negative finding. Acetaminophen did not lower the mean bodytemperature during the fever episodes associated with a FS recurrence, although it wasantipyretic in episodes without seizures (IV). Similarly no antipyretic effect of ibuprofenduring fever episodes with seizures was observed by van Stuijvenberg et al. (1998).Likewise Schneiderman et al. (1993) reported a lack of antipyretic effect in the case ofhigher doses of acetaminophen (90-120 mg/kg/24h) among children hospitalized for FS.

41

Our systematic review of PGs and their synthetase inhibitors showed that NSAIDs inadult animals did not inhibit MES-induced seizures, which are a prototype model forgeneralized tonic-clonic seizures (Fisher 1993) and thus the best of the current animalmodels for FS (I). This finding supports the assumption that NSAIDs might not have anyprophylactic effect on FS at any dose, although no direct conclusions concerning FS canbe drawn from animal studies (I).

The clinical failure of PG synthetase inhibitors to prevent recurrences is interesting. Itsuggests that either PGs are not important in the pathogenesis of FS or the diversity ofinhibition mechanisms in PG synthesis nullifies the effect. If this is the case, trials usingdifferent types of NSAIDs are urgently warranted, as they could lead to the developmentof selective NSAID preparations capable of improving seizure inhibition andneuroprotection.

Even though acetaminophen, a weak inhibitor of PG synthesis, does not seem toprevent FS, it is unlikely to counteract the synthesis of seizure-inhibiting PGs. Thus itmay be the most suitable antipyretic medication for children with FS.

6.5 Anticonvulsant treatment for preventing recurrences

Both phenobarbital and valproate were effective in preventing recurrences in our meta-analysis, but these drugs are not recommended for FS patients on account of their side-effects (Committee on Drugs 1982, Farwell et al. 1990).

The use of intermittent diazepam has been advocated since the observation byKnudsen and Vestermark (1978) that the effect of intermittent diazepam suppositories onrecurrences was similar to that of continuous phenobarbital treatment. The meta-analysisnevertheless found only three placebo-controlled trials evaluating the effect ofintermittent diazepam. In our treatment trial, diazepam administered during fever did notprevent recurrences in unselected FS patients (IV). The low dosage of diazepam used mayhave contributed to this negative result, but the same dose has previously been reported tobe effective in preventing recurrences (Knudsen 1985). Rosman et al. (1993) showed ahigher dose of diazepam (0.33 mg/kg every eight hours) to have a prophylactic effect inchildren with an increased risk of recurrences in a placebo-controlled setting, theintention-to-treat analysis revealing a relative risk of 0.56 (95% CI 0.38-0.81, p = 0.002)for recurrences in children assigned to diazepam compared with the placebo group. Theeffectiveness of diazepam in this case may have been due to the dose, but also to patientselection, as about a half of the patients were participating in the trial after their firstrecurrence. The third placebo-controlled trial led to the conclusion that diazepam had noeffect on recurrences (Autret et al. 1990). Our meta-analysis combining all the availabledata shows that intermittent use of diazepam appears to be ineffective for preventing FSrecurrences.

If diazepam is recommended as prophylaxis for FS, the dose should be 0.33 mg/kgevery eight hours while fever last. Its side-effects at high doses can make the follow-up ofa feverish child difficult and also cause problems with compliance. As the prognosis forFS is benign (Knudsen et al. 1996), the use of intermittent diazepam should be carefullyconsidered.

42

6.6 Febrile seizures and mesial temporal sclerosis

Our finding that patients with different forms of FS had no signs of MTS indicates that itsoccurrence after a FS is an uncommon event. Due to the relatively small sample ofchildren with very prolonged FSs, their role as a risk factor for MTS could not beexcluded with certainty.

An average right-left absolute hippocampal volume difference of 0.2 cm3 has beenreported in healthy adults (Jack et al. 1990). Using standard deviations from this series,lower and upper threshold values of –0.2 cm3 and 0.6 cm3 were calculated and then usedto classify the patients into groups with right-sided hippocampal atrophy, nonlateralizingvolume difference and left-sided hippocampal atrophy (Jack et al. 1992). Since there areno normal values for adolescent patients, we used the findings in our control group, i.e.patients with a single simple FS, as a source for reference values. Our finding of a meanright-left hippocampal volume difference of 0.18 cm3 is quite close to the above value of0.2 cm3 reported for adults. The right-left hippocampal volume difference thresholdvalues in our adolescent control patient group (2 SD range) were –0.13m3 and 0.49 cm3.

Our results demonstrate a significantly smaller hippocampal right–left volumedifference, without any decrease in mean absolute hippocampal volumes, among thepatients with a prolonged initial FS than among the matched controls. Right-left volumeasymmetry in patients with severe TLE and unilateral hippocampal sclerosis haspreviously been linked to hippocampal atrophy (Kuks et al. 1993, Free et al. 1996), butwe could not find this in our patients. Perhaps there are subtle forms of MTS that cannotbe detected with MRI, although this has been considered to be the most sensitive non-invasive method for evaluating moderate and severe MTS (Jack et al. 1992). Only furtherfollow-up of the patients can reveal whether the cases of hippocampal volume asymmetrywithout signs of absolute volume reduction and structural damage will develop intounilateral hippocampal atrophy. Our present follow-up period was nevertheless a longone, so that if changes were to develop as a consequence of FS the findings should havebeen in evidence by now.

Conclusions concerning the outcome of FS cannot be derived from highly selectedTLE patient series, for in cases of severe TLE the fever may have been the first triggerleading to the seizures and the structural changes in the mesial temporal area may wellhave been there earlier (Fernandez et al. 1998).

MTS is found in only about one per cent of children with newly diagnosed epilepsy(King et al. 1998, Berg et al. 2000), but its occurrence in children with FS has notpreviously been evaluated. We collected a population-based series of patients andfollowed them up prospectively from the first FS, selecting the ones most likely to havean adverse sequel, i.e. those with a prolonged initial FS or unprovoked recurrences. It isnot probable that new cases with epilepsy will turn up later among these FS patients,because of the long follow-up period. If epilepsy arises, usually it occurs within nineyears of the FS (Annegers et al. 1979), and our follow-up time was over 12 years.Although the sample size was relatively small, our results support the conclusion thatthere is no causal relation between FSs and MTS in an unselected series of FS patients.

7 ConclusionsElevated amounts of PGE2 and PGF2alfa occur in the CSF after a FS. Our systematicreview demonstrated that different endogenous PGs have seizure-modulating effectsranging from inhibition (PGD2, PGE1 and PGE2) to provocation (PGF2alfa) in animals(I). Thus the PG system is probably important in the pathogenesis of FS.

High fever and frequent fever episodes are powerful risk factors for recurrences (II).Efforts to prevent frequent febrile infections by restricting the number of infectiouscontacts, e.g. by rearranging day-care, might reduce the risk of recurrence, but this theoryhas not yet been proved.

Acetaminophen did not have an antipyretic effect during fever episodes withrecurrences, and it did not prevent FS in our treatment trial (IV). There is no evidence thatFS recurrences could be prevented with antipyretic medication. PG synthetase inhibitors(NSAIDs) attenuate fever, but as they modulate the PG balance in the brain, they couldsimultaneously alter seizure susceptibility. The seizure-modulating effects of NSAIDswere evident in our review, but diverse results had been obtained in experimental modelsinvolving adult animals. These findings have not been confirmed in humans (I).

Low intermittent doses of diazepam did not reduce recurrences of FS in our treatmenttrial (IV). Similarly, our meta-analysis of prophylactic treatment against FS revealed theineffectiveness of diazepam for preventing recurrences (III). Even though bothcontinuous phenobarbital and valproate have been shown to reduce recurrences, their useis not acceptable on account of their severe side-effects. Thus there is no safe andeffective medication available for the prevention of FS.

We found no MTS in our prospectively followed FS patients (V). Its occurrence, evenfollowing a prolonged FS, must be an uncommon event, confirming the good clinicaloutcome of FS.

8 ReferencesAkarasu ES, Mamuk S & Comert A (1998) Inhibition of pentylenetetrazol-induced seizures in rats

by prostaglandinD2. Epilepsy Res 30: 63-68.Alcami A & Smith GI (1996) A mechanism for the inhibition of fever by virus. Proc Nat Acad Sci

USA 93: 11029-11034. American Academy of Pediatrics (1999) Practice parameter: Long-term treatment of the child with

simple febrile seizures. Pediatrics 103: 1307-1309.Annegers JF, Hauser WA, Elveback LR & Kurland LT (1979) The risk of epilepsy following febrile

convulsions. Neurology 29: 297-303.Antony JH & Hawke SHB (1983) Phenobarbital compared with carbamazepine in prevention of re-

current febrile convulsions. Am J Dis Child 137: 892-895.Autret E, Billard C, Bertrand P, Motte J, Pouplard F & Jonville AP (1990) Double-blind, randomized

trial of diazepam versus placebo for prevention of recurrence of febrile seizures. J Pediatr 117:490-494.

Bacon CJ, Hierons AM, Mucklow JC, Webb J, Rawlins MD & Weightman D (1981) Placebo con-trolled study of phenobarbitone and phenytoin in the prophylaxis of febrile convulsions. Lancet 2:600-604.

Baik EJ, Kim EJ, Lee SH & Moon C (1999) Cyclo-oxygenase-2 selective inhibitors aggravate kainicacid induced seizure and neuronal cell death in hippocampus. Brain Res 843: 118-129.

Barone SR, Kaplan MH & Krilov LR (1995) Human herpesvirus-6 infection in children with first fe-brile seizures. J Pediatr 127: 95-97.

Barr WB, Ashtari M & Schaul N (1997) Bilateral reductions in hippocampal volume in adults withepilepsy and a history of febrile seizures. J Neurol Neurosurg Psychiatry 63: 461-467.

Bauman RJ & Duffner PK (2000) Treatment of children with simple febrile seizures: the AAP prac-tice parameter. Pediatr Neurol 23: 11-17.

Baumer JH, David TJ, Valentine SJ, Roberts JE & Hughes BR (1981) Many parents think their childis dying when having a first febrile convulsion. Dev Med Child Neurol 23: 462-464.

Berg AT, Shinnar S, Hauser WA & Leventhal J (1990) Predictors of recurrent febrile seizures: Ameta-analytic review. J Pediatr 116: 329-337.

Berg AT, Shinnar S & Hauser WA (1992) A prospective study of recurrent febrile seizures. N EnglJ Med 327: 1122-1127.

Berg AT, Shinnar S, Shapiro ED, Salomon ME, Crain EF & Hauser WA (1995) Risk factors for afirst febrile seizure: matched case-control study. Epilepsia 36: 334-341.

Berg AT & Shinnar S (1996) Unprovoked seizures in children with febrile seizures: short-term out-come. Neurology 47: 562-568.

45

Berg AT, Testa FM, Levy SR & Shinnar S (2000) Neuroimaging in children with newly diagnosedepilepsy: a community-based study. Pediatrics 106: 527-532.

Berti F, Bernareggi V, Folco GC, Fumagalli R & Paoletti R (1976) ProstaglandinE2 and cyclic nu-cleotides in rat convulsions and tremors. Adv Biochem Psychopharmacol 15: 367-377.

Bhattacharya SK & Sanyal AK (1978) Inhibition of pentylenetetrazol-induced convulsions in rats byprostaglandinE1: role of brain monoamines. Psychopharmacology 56: 235-237.

Bhattacharya SK & Parmar SS (1987) ProstaglandinD2 inhibits pentylenetetrazol-induced convul-sions in rats by a serotonin-mediated mechanism. Pharm Res 4: 406-408.

Blatteis CM & Sehic E (1997) Circulating pyrogen signaling of the brain. Annals New York Acade-my of Sciences 813: 445-447.

Bodenza A & Wisniewski K (1977) The effect of prostaglandinE1 on central cholinergic mecha-nisms. Pharmacology 15: 143-151.

Callard RE & Gearing AJH (1994) The cytokine facts book. Academic Press, London, p 1-30.Camfield PR, Camfield CS, Shapiro SH & Cummings C (1980) The first febrile seizure: antipyretic

instruction plus either phenobarbital or placebo to prevent recurrence. J Pediatr 97: 16-21.Cendes F, Andermann F, Dubeau F, Gloor P, Evans A, Jones-Gotman M, Olivier A, Andermann E,

Robitaille Y, Lopes-Cendes I et al (1993) Early childhood prolonged febrile convulsions, atrophyand sclerosis of mesial structures, and temporal lobe epilepsy - an MRI volumetric study. Neurol-ogy 43: 1083-1087.

Chen K, Baram TZ & Soltesz I (1999) Febrile seizures in the developing brain result in persistentmodification of neuronal excitability in limbic circuits. Nat Med 5: 888-894.

Climax J & Sewell RD (1981) Modification of convulsive behaviour and body temperature in miceby intracerebroventricular administration of prostaglandins, arachidonic acid and the soluble ace-tylsalicylic acid salt lysine acetylsalicylate. Arch Int Pharmacodyn Ther 250: 254-265.

Committee on Drugs (1982) Valproic acid: benefits and risks. Pediatrics 70: 316-319. Consensus Development Conference on Febrile Seizures (1980) Pediatrics 66: 1009-1012.Croucher MJ, Marriott DR, Bradford HF & Wilkin GP (1991) Lack of effect of focally administered

prostaglandins on electrically kindled seizure activity. Prostaglandins 42: 29-38.Duru S & Turker RK (1969) Effect of prostaglandinE1on the strychnine-induced convulsion in the

mouse. Experientia 25: 275. Ellenberg JH & Nelson KB (1978) Febrile seizures and later intellectual performance. Arch Neurol

35: 17-21.El-Radhi AS & Banajeh S (1989) Effect of fever on recurrence rate of febrile convulsions. Arch dis

Child 64: 869-870.Faden A, Golbus M & Spire JP (1976) Electroencephalographic changes following intra-amniotic

prostaglandin F2 alpha administration for therapeutic abortion. Obstet Gynecol 47: 607-608.Faero O, Kastrup KW, Lykkegaard Nielsen E, Melchior JC & Thorn I (1972) Successful prophylaxis

of febrile convulsions with phenobarbital. Epilepsia 13: 279-285.Falconer MA (1974) Mesial temporal (Ammon's horn) sclerosis as a common cause of epilepsy. Ae-

tiology, treatment, and prevention. Lancet 28: 767-770.Farwell JR, Lee YJ, Hirtz DG, Sulzbacher SI, Ellenberg JH & Nelson KB (1990) Phenobarbital for

febrile seizures - effects on intelligence and on seizure recurrence. N Engl J Med 322: 364-369.Fernandez G, Effenberger O, Vinz B, Steinlein O, Elger CE, Döhring W & Heinze HJ (1998) Hip-

pocampal malformation as a cause of familial febrile convulsions and subsequent hippocampalsclerosis. Neurology 50: 909-917.

Fisher RS (1993) Emerging antiepileptic drugs. Neurology 43: 12-20.Forsgren L, Sidenvall R, Blomquist HK, Heijbel J & Nystrom L (1990a) An incident case-referent

study of febrile convulsions in children: genetical and social aspects. Neuropediatrics 21: 153-159.

46

Forsgren L, Sidenvall R, Blomkvist HK & Heijbel J (1990b) A prospective incidence study of febrileconvulsions. Acta Paediatr Scand 79: 550-557.

Förstermann U, Heldt R, Knappen F & Hertting G (1982) Potential anticonvulsive properties of en-dogenous prostaglandins formed in mouse brain. Brain Res 240: 303-310.

Förstermann U, Heldt R & Hertting G (1983) Effects of intracerebroventricular administration ofprostaglandinD2 on behaviour, blood pressure and body temperature as compared toprostaglandinE2 and F2alpha. Psychopharmacology (Berlin) 80: 365-370.

Förstermann U, Seregi A & Hertting G (1984) Anticonvulsive effects of endogenous prostaglandinsformed in brain of spontaneously convulsing gerbils. Prostaglandins 27: 913-923.

Fraser IS & Gray C (1974) Letter: electroencephalogram and changes after prostaglandin. Lancet 1:360.

Free SL, Li LM, Fish DR, Shorvon SD & Stevens JM (1996) Bilateral hippocampal volume loss inpatients with a history of encephalitis or meningitis. Epilepsia 37: 400-405.

Harvey AS, Grattan-Smith JD, Desmond PM, Chow CW & Berkovic SF (1995) Febrile seizures andhippocampal sclerosis: Frequent and related findings in intractable temporal lobe epilepsy ofchildhood. Pediatr Neurol 12: 201-206.

Hedges LV. Statistical considerations. In: Cooper H, Hedges LV (1994) The handbook of researchsynthesis. New York: Russel Sage Foundation p 29-38.

Helminen M & Vesikari T (1990) Increased interleukin-1 production from LPS-stimulate peripheralblood monocytes in children with febrile convulsions. Acta Paediatr Scand 79: 810-816.

Hukin J, Farrell K, MacWilliam LM, Colbourne M, Waida E, Tan R, Morz L & Thomas E (1998)Case-control study of primary human herpesvirus-6 infection children with febrile seizures. Pedi-atrics 101(2): E3.

Ichyama T, Nishikawa M, Yoshitomi T, Hayashi T & Furukawa S (1998) Tumor necrosis factor al-pha, interleukin-1beeta and interleukin-6 in cerebrospinal fluid from children with prolonged fe-brile seizures. Neurology 50: 407-411.

Jack CR, Sharbrough FW, Twomey CK, Cascino GD, Hirschorn KA, Marsh WR, Zinsmeister AR &Scheithauer B (1990) Temporal lobe seizures: lateralization with MR volume measurements ofthe hippocampal formation. Radiology 175: 423-429.

Jack CR, Sharbrough FW, Cascino GD, Hirschorn KA, O’Brian PC & Marsh WR (1992) Magneticresonance image-based hippocampal volumetry: correlation with outcome after temporal lobecto-my. Ann Neurol 31:138-146.

Jee SH, Long CE, Schnabel KC, Seghal N, Epstein LG & Hall CB (1998) Risk of recurrent seizuresafter a primary human herpesvirus-6 induced febrile seizure. Pediatr Infect Dis J 17: 43-48.

King MA, Newton MR, Jackson GD, Fitt GJ, Mitchell LA & Silvapulle MJ (1998) Epileptology ofthe first-seizure presentation: a clinical, electroencephalographic, and magnetic resonance imag-ing study of 300 consecutive patients. Lancet 352: 1007-1011.

Knudsen FU & Vestermark S (1978) Prophylactic diazepam or phenobarbitone in febrile convul-sions: a prospective, controlled study. Arch Dis Child 53: 660-663.

Knudsen FU (1985) Recurrence risk after first febrile seizure and effect of short term diazepamprophylaxis. Arch Dis Child 60: 1045-1049.

Knudsen FU (1988) Frequent febrile episodes and recurrent febrile convulsions. Acta Neurol Scand78: 414-417.

Knudsen FU (1990) Febrile Convulsions. Comprehensive Epileptology . New York, Raven Press, p133-143.

Knudsen FU, Paerregaard A, Andersen R & Andersen J (1996) Long term outcome of prophylaxisfor febrile convulsions. Arch Dis Child 74: 13-18.

Kondo K, Nagafuji H, Hata A, Tomomori C & Yamanishi K (1993) Association of human herpesvi-rus-6 infection of the central nervous system with recurrence of febrile convulsions. J Infect Dis167: 1197-1200.

47

Kuks J, Cook M, Fish D, Stevens J & Shorvon S (1993) Hippocampal sclerosis in epilepsy and child-hood febrile seizures. Lancet 342: 1391-1394.

Lawson J, Cook M, Bleasek A, Nayanar V, Morris K & Bye A (1997) Quantitative MRI in outpatientchildhood epilepsy. Epilepsia 38: 1289-1293.

Lewis HM, Parry JV, Parry RP, Davies HA, Sanderson PJ, Tyrell DAJ & Valman HB (1979) Role ofviruses in febrile convulsions. Arch Dis Child 54: 869-876.

Licinio J & Wong ML (1996) Interleukin-1beeta and fever. Nature Medicine 2: 1314-1315. Louhiala PJ, Jaakkola N, Ruotsalainen R, Jaakkola JJK (1995) Form of day care and respiratory in-

fections among Finnish children. Am J Public Health 85: 1109-1112.Lyneham R, Low P, McLeod J, Shearman R, Smith I & Korda A (1973) Convulsions and electroen-

cephalogram abnormalities after intra-amniotic prostaglandinF2alfa. Lancet 2: 1003-1005.Löscher W & Siemes H (1988) Increased concentration of prostaglandinE-2 in cerebrospinal fluid of

children with febrile convulsions. Epilepsia 29: 307-310.MacKenzie I, Hillier K & Embrey M (1973) Letter: convulsions and prostaglandin-induced abortion.

Lancet 2: 1323.Mackintosh (1970) Studies on prophylactic treatment of febrile convulsions in children. Is it feasible

to inhibit attacks by giving drugs at the first sign of fever or infection? Clin Pediatr 9: 283-286. Madan BR, Gupta RS & Madan V (1974) Actions of protaglandinsE1, E2, F1alpha and F2alpha in

ouabain-induced arrhythmia and maximal electroshock induced seizures. Indian J Med Res 62:1647-1651.

Mamelle N, Mamelle JC, Plasse JC, Revol M & Gilly R (1984) Prevention of recurrent febrile con-vulsions - a randomised therapeutic assay: sodium valproate, phenobarbital and placebo. Neuro-pediatrics 15: 37-42.

Marsh L, Morrell M, Shear P, Sullivan E, Freeman H, Marie A, Lim K & Pfefferbaum A (1997) Cor-tical and hippocampal volume deficits in temporal lobe epilepsy. Epilepsia 38: 576-587.

McGinley S, Centra M & Lysz TW (1985) The effect of inhibiting brain thromboxane biosynthesison pentylenetetrazole-induced seizure threshold. J Neurosci Res 13: 563-567.

McKiernan J, Mellor DH & Court S (1981) A controlled trial of pyridoxine supplementation in chil-dren with febrile convulsions. Clin Pediatr 20: 208-211.

Millichap JG (1959) Studies in febrile seizures. I Height of body temperature as a measure of the fe-brile seizure threshold. Pediatrics 23: 76-85.

Navarro E, Romero SD & Yaksh TL (1989) CNS stimulation and PGE2 release. III Pentamethylene-tetrazole-induced seizures. J Cereb Blood Flow Metab 9: 180-186.

Nelson KB & Ellenberg JH (1976) Predictors of epilepsy in children who have experienced febrileseizures. N Engl J Med 295: 1029-1033.

Nelson KB & Ellenberg JH (1978) Prognosis in children with febrile seizures. Pediatrics 61: 720-726.Newton RW (1988) Randomized controlled trials of phenobarbitone and valproate in febrile convul-

sions. Arch Dis Child 63: 1189-1191.Offringa M, Derksen-Lubsen G, Bossuyt PM & Lubsen J (1992) Seizure recurrence after a first fe-

brile seizure: a multivariate approach. Dev Med Child Neurol 34: 15-24.Rantala H, Uhari M & Tuokko H (1990) Viral infections and recurrences of febrile convulsions. J

Pediatr 116: 195-199. Rantala H & Uhari M (1994) Risk factors for recurrences of febrile convulsions. Acta Neurol Scand

90: 207-210.Rantala H, Uhari M & Hietala J (1995) Factors triggering the first febrile seizure. Acta Paediatr 84:

407-410.Rosenkranz RP (1978) Effects of intracerebroventricular administration of PGE1, E2 and F2alpha on

electrically induced seizures in mice. Prostaglandins 15: 925-942.

48

Rosenkranz RP & Killam KF Jr (1979) Effects of intracerebroventricular administration ofprostaglandinsE1 and E2 on chemically induced convulsions in mice. J Pharmacol Exp Ther 209:231-237.

Rosenkranz RP & Killam KF Jr (1981) Anticonvulsant effects of PGE2 on electrical, chemical andphotomyoclonic animal models of epilepsy. Prog Lipid Res 20: 515-522.

Rosman NP, Colton T, Labazzo J, Gilbert PL, Gardella NB, Kay EM, Van Bennekom C & WinterMR (1993) A controlled trial of diazepam administered during febrile illnesses to prevent recur-rence of febrile seizures. N Engl J Med 329: 79-84.

Salmenperä T, Kälviäinen R, Partanen K & Pitkänen A (1998) Hippocampal damage caused by sei-zures in temporal lobe epilepsy. Lancet 351: 35.

Saper CB & Breder CD (1994) The neurologic basis of fever. New Engl J Med 330: 1880-1886.Schnaiderman D, Lahat E, Sheefer T & Aladjem M (1993) Antipyretic effectiveness of acetami-

nophen in febrile seizures: ongoing prophylaxis versus sporadic usage. Eur J Pediatr 152: 747-749.

Seregi A, Folly G, Antal M, Serfozo P & Schaefer A (1981) Studies on prostaglandinF2alfa formationcaused by pentametylenetetrazol-induced convulsions in rat brain. Prostaglandins 21: 217-226.

Seregi A, Förstermann U & Hertting G (1984) Decreased levels of brain cyclo-oxygenase productsas a possible cause of increased seizure susceptibility in convulsion-prone gerbils. Brain Res 305:393-395.

Seregi A, Förstermann U, Heldt R & Hertting G (1985) The formation and regional distribution ofprostaglandinsD2 and F2alpha in the brain of spontaneously convulsing gerbils. Brain Res 337:171-174.

Shearman R, Lyneham R, Walsh J, Itzkowic D & Shutt D (1975) Electroencephalographic changesafter intra-amnioitic prostaglandinF2alfa and hypertonic saline. Br J Obstet Gynaecol 82: 314-317.

Simmet T & Tippler B (1990) Cysteinyl-leukotriene production during limbic seizures trigged by kai-nic acid. Brain Res 515: 79-86.

Steinhauer HB, Anhut H & Hertting G (1979) The synthesis of prostaglandins and thromboxane inthe mouse brain in vivo. Influence of drug induced convulsions, hypoxia and anticonvulsants tri-methadione and diazepam. Naunyn Schmiedeberg’s Arch Pharmacol 310: 53-58.

Steinhauer HB & Hertting G (1981) Lowering the convulsive threshold by non-steroidal anti-inflam-matory drugs. Eur J Pharmacol 69: 199-203.

Tamai I, Takei T, Maekawa K & Ohta H (1983) Prostaglandin F2alfa concentrations in the cerebro-spinal fluid of children with febrile convulsions, epilepsy and meningitis. Brain Dev 5: 357-362.

Tanner JM & Whitehouse RH (1976) Clinical longitudinal standards for height, weight, height veloc-ity and stages of puberty. Arch Dis Child 51: 170-182.

Theodor W, Bhatia S, Hatta J, Fazilat S, DeCarli C, Bookheimer S & Gaillard W (1999) Hippocampalatrophy, epilepsy duration and febrile seizures in patients with partial seizures. Neurology 52:132-136.

Thiery M, Amy J, De Hemptinne D, Le Sian A (1974) Letter: prostaglandins and convulsions. Lancet1: 218.

Thompson SG (1994) Why sources of heterogeneity in meta-analysis should be investigated. BMJ309: 1351-1355.

Trenerry M, Jack C, Sharbrough F, Cascino G, Hirschorn K, Marsh W, Kelly P & Meyer F (1993)Quantitative MRI hippocampal volumes: association with onset and duration of epilepsy, and fe-brile convulsions in temporal lobectomy patients. Epilepsy Res 15: 247-252.

Tutuncuoglu S, Kutukculer N, Kepe L, Coker C, Berdeli A & Tekgul H (2001) Proinflammatory cy-tokines, prostaglandins and zinc in febrile convulsions. Pediatrics International 43: 235-239.

Uhari M & Möttönen M (1999) An open randomised controlled trial of infection prevention in childday-care centers. Pediatr Infect Dis J 18: 672-677.

49

Ushikubi F, Segi E, Sugimoto Y, Murata T, Matsuoka T, Kobayashi T, Hizaki H, Tuboi K, Kat-suyama M, Ichikawa A et al (1998) Impaired febrile response in mice lacking the prostaglandinE receptor subtype EP3. Nature 395: 281-284.

Vane JR, Botting RM & Botting J (1996) Improved non-steroid anti-inflammatory drugs COX-2 en-zyme inhibitors. Kluwer Academic Publishers BV, Dordrecht, The Netherlands, p. 1-27.

VanLandingham K, Heinz E, Cavazos J & Lewis D (1998) Magnetic resonance imaging evidence ofhippocampal injury after prolonged focal febrile convulsions. Ann Neurol 43: 413-426.

Van Stuijvenberg M, Derksen-Lubsen G, Steyerberg EW, Habbema JDF & Moll HA (1998) Rand-omized, controlled trial of ibuprofen syrup administered during febrile illnesses to prevent febrileseizure recurrences. Pediatrics 102: E511-517.

Verity CM, Butler NR & Golding J (1985a) Febrile convulsions in a national cohort followed up frombirth. I - Prevalence and recurrences in the first five years of life. BMJ 290: 1307-1310.

Verity CM, Butler NR & Golding J (1985b) Febrile convulsions in a national cohort followed up frombirth. II – Medical history and intellectual ability at 5 years. BMJ 290: 1311-1315.

Verity CM & Golding J (1991) Risk of epilepsy after febrile convulsion: a national cohort study. BMJ303: 1373-1376.

Verity CM, Greenwood R & Golding J (1998) Long term intellectual and behavioural outcomes ofchildren with febrile convulsions. N Engl J Med 338: 1723-1728.

Wallenstein MC & Mauss EA (1984) Effect of prostaglandin synthetase inhibitors on experimentallyinduced convulsions in rats. Pharmacology 29: 85-93.

Wallenstein MC (1985) Differential effect of prostaglandin synthetase inhibitor pretreatment onpentylenetetrazol –induced seizures in rat. Arch Int Pharmacodyn: 275: 93-104.

Wallenstein MC (1987) Attenuation of penicillin models of epilepsy by non-steroidal anti-inflamma-tory drugs. Exp neurol 98: 152-160.

Wallenstein MC (1991) Attenuation of epileptogenesis by non-steroidal anti-inflammatory drugs inthe rat. Neuropharmacology 30: 657-663.

Watson C, Andermann F, Gloor P, Jones-Gotman M, Peters T, Evans A, Olivier A, Melanson D &Leroux G (1992) Anatomic basis of amygdaloid and hippocampal volume measurement by mag-netic resonance imaging. Neurology 42: 1743-1750.

Wolfe L & Mamer O (1975) Measurement of prostaglandinF2alfa levels in human cerebrospinal fluidin normal and pathological conditions. Prostaglandins 9: 183-192.