Initiation and Discontinuation of Antiepileptic Drugs

Varda Gross Tsur

Christine O’Dell

Shlomo Shinnar

Supported in part by grant 1 R01 NS26151 (S. Shinnar) from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland.

Portions of this chapter are reprinted from O’Dell C, Shinnar S. Initiation and discontinuation of antiepileptic drugs. Neurol Clin 2001;19:289-311, with permission.

Over the past two decades there has been much information about the prognosis of seizure disorders, the effects of antiepileptic drug (AED) therapy on prognosis, and the relative risks of both seizures and of AED therapy. This chapter reviews the clinical decision making in initiating and discontinuing AEDs in children and adults, with particular emphasis on the data regarding the recurrence risk for seizures in different settings and the effect of AEDs on this risk. The risks and benefits of initiating and discontinuing AED therapy are then addressed in the context of an individualized therapeutic approach which emphasizes weighing the risks and benefits of drug therapy versus both the statistical risk of another seizure and the consequences of such an event.

RECURRENCE RISK FOLLOWING A FIRST UNPROVOKED SEIZURE

To develop a rational approach to the management of individuals who present with an initial unprovoked seizure, it is necessary to have some understanding of the natural history and prognosis of the disorder in this setting. Approximately one-third to one-half of children and adults with seizures will initially present to medical attention following a single seizure (1,2). The remainder will already have a history of prior events at the time of presentation. It is the group who present with a single seizure that are most relevant to this discussion. In accordance with the International League Against Epilepsy (ILAE) guidelines for epidemiologic research in epilepsy, a first unprovoked seizure is defined as a seizure or flurry of seizures all occurring within 24 hours in a person older than 1 month of age with no prior history of unprovoked seizures (3).

Since 1982, a number of studies have attempted to address the recurrence risk following a first unprovoked seizure using a variety of recruitment and identification techniques (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22). The reported overall recurrence risk following a first unprovoked seizure in children and adults varies from 27% to 71%. Studies that carefully excluded those with prior seizures report recurrence risks of 27% to 52% (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18). The higher recurrence risks are, with one exception (19), reported from studies that included subjects who already had recurrent seizures at the time of identification and who were thus more properly considered to have newly diagnosed epilepsy.

While there is considerable disparity in the absolute recurrence risk reported in the different studies, the time course of recurrence is remarkably similar among all studies (5). The majority of recurrences occur early, with approximately 50% of recurrences occurring within 6 months of the initial seizure and over 80% within 2 years of the initial seizure (5,13). Late recurrences are unusual, but they have occurred up to 10 years after the initial seizure (13,14). This time course is true both in studies that report low and high recurrence risks (4,5,7, 8, 9, 10,12, 13, 14,19, 20, 21).

A relatively small number of factors are associated with a differential recurrence risk. The most important of these are the etiology of the seizure, the electroencephalogram (EEG), and whether the first seizure occurred in wakefulness or sleep. These factors are consistent across most studies regardless of the absolute risk of recurrence reported in the individual study (4,5,7, 8, 9, 10, 11, 12, 13, 14, 15,18,20,21). Factors not associated with a significant change in the recurrence risk include age of onset, the number of seizures in the first 24 hours, and the duration of the initial seizure. The absolute recurrence risks appear similar in children and adults (5), although the consequences of such a recurrence are quite different. Selected risk factors are discussed below.

Etiology

In the ILAE classification, etiology of seizures is classified as remote symptomatic, cryptogenic, or idiopathic (3). Remote symptomatic seizures are those without an immediate cause but with an identifiable prior brain injury or the presence of a static encephalopathy such as mental retardation or cerebral palsy, which are known to be associated with an increased risk of seizures. Cryptogenic seizures are those occurring in otherwise normal individuals with no clear etiology. Until recently, cryptogenic seizures were also called idiopathic. In the new classification, idiopathic is reserved for seizures occurring in the context of the presumed genetic epilepsies such as benign rolandic and childhood absence (23,24). However, much of the literature on the recurrence risk following a first unprovoked seizure lumps idiopathic and cryptogenic together as idiopathic using the original classification developed by Hauser and coworkers (8).

Not surprisingly, both children and adults with a remote symptomatic first seizure have higher risk of recurrence than those with a cryptogenic first seizure. A meta-analysis of the studies published up to 1990 found that the relative risk of recurrence following a remote symptomatic first seizure was 1.8 (95% confidence interval, 1.5, 2.1) compared to those with a cryptogenic first seizure (5). Comparable findings are reported in more recent studies (13,15,21). Idiopathic first unprovoked seizures occur almost exclusively in children. Although the long-term prognosis of these children is quite favorable, the recurrence risk is actually comparable to those with a remote symptomatic first seizure (13). This is because, by definition, to meet the criteria for an idiopathic first seizure, they must have an abnormal EEG (23,24).

Electroencephalogram

The EEG is an important predictor of recurrence, particularly in cases that are not remote symptomatic and in children (5,7,8,10, 11, 12, 13,15, 16, 17, 18,21,25). Studies of recurrence risk following a first seizure in childhood have uniformly reported that those with an abnormal EEG have a higher recurrence risk than those with a normal EEG (5,7,12,13,15,21,25). For this reason, the American Academy of Neurology’s recently published guideline on the evaluation of children with a first unprovoked seizure considers an EEG to be a standard part of the evaluation (21). Epileptiform abnormalities are more important than nonepileptiform ones, but any EEG abnormality increases the recurrence risk in cases that are not remote symptomatic (25). In our study, the risk of seizure recurrence within 24 months for children with an idiopathic/cryptogenic first seizure was 25% for those with a normal EEG, 34% for those with nonepileptiform abnormalities, and 54% for those with epileptiform abnormalities (25). Whereas in our data, any clearly abnormal electroencephalographic patterns, including generalized spike and wave, focal spikes, and focal or generalized slowing, increased the risk of recurrence, Camfield and associates (7) reported that only epileptiform abnormalities substantially increase the risk of recurrence in children. Despite minor disagreements as to which electroencephalographic patterns are most significant, the EEG appears to be the most important predictor of recurrence in children with a cryptogenic/idiopathic first seizure. In addition, it is the EEG that primarily distinguishes whether a neurologically normal child with a first seizure is classified as cryptogenic or idiopathic.

In adults, the data are more controversial. The majority of studies do find an increased recurrence risk associated with an abnormal EEG (5,9,10,18), although one study failed to find a significant effect (11). Hauser and colleagues (8) found that generalized spike-and-wave patterns are predictive of recurrence but not focal spikes. A meta-analysis of these studies concluded that the overall data do support an association between an abnormal EEG and an increased recurrence risk in adults as well (5), although which electroencephalographic patterns besides generalized spike and wave are important remains unclear (5,9,10,18).

Sleep State at Time of First Seizure

In adults, seizures that occur at night are associated with a higher recurrence risk than those that occur in the day-time (11). In children, whose sleep patterns may include day-time naps, the association is more clearly between sleep state and recurrence risk rather than time of day (13,26). Interestingly, the association is not just because nocturnal seizures tend to occur in certain epilepsy syndromes. Thus, even children whose EEG has centrotemporal spikes and who meet the criteria for benign rolandic seizures (24) have a higher recurrence risk if the first seizure occurs during sleep than if it occurs while awake (26). Furthermore, if the first seizure occurs during sleep, there is a high likelihood that the second one, should it occur, will also occur during sleep (26). In our series, the 2-year recurrence risk was 53% for children whose initial seizure occurred during sleep compared with a 30% risk for those whose initial
seizure occurred while awake (13). On multivariable analysis, etiology, the EEG, and sleep state were the major significant predictors of outcome. From a therapeutic point of view, the implication of a seizure during sleep is unclear. While the recurrence risk is higher, recurrences will tend to occur in sleep. As the major risk of a brief seizure in children or adults is that it may happen at a time or place where the impairment of consciousness will have serious consequences, the morbidity of a seizure during sleep is fairly low in both cases.

Seizure Classification

In some studies, the risk of recurrence following a first unprovoked seizure is higher in subjects with a partial seizure than in those with a generalized first seizure (5). This association is mostly found on univariate analysis and disappears once the effect of etiology and the EEG are accounted for (5,8,12,13). Partial seizures are more common in those with a remote symptomatic first seizure and in children with an abnormal EEG (12). Note that some generalized seizure types, such as absence and myoclonic, very rarely present as a first seizure and so would be excluded from studies of first seizure (16,21). Generalized seizures that present to medical attention at the time of the first seizure are usually tonic-clonic (13).

Duration of Initial Seizure

In children, the duration of the first seizure is not associated with a differential recurrence risk. In our study, 48 (12%) of 407 children (38 cryptogenic/idiopathic, 10 remote symptomatic) presented with status epilepticus (duration longer than 30 minutes) as their first unprovoked seizure (13). The recurrence risk in these children was not different than in children whose first seizure was briefer. However, if a recurrence did occur it was likely to be prolonged (13,27). Of the 24 children with an initial episode of status who experienced a seizure recurrence, 5 (21%) recurred with status. Of the 147 children who presented with an initial brief seizure and experienced a seizure recurrence, only 2 (1%) recurred with status epilepticus (p <0.001). In adults there is a suggestion that a prolonged first seizure, particularly in remote symptomatic cases, is associated with a higher risk of recurrence (10).

Treatment Following a First Seizure

Four randomized clinical trials in children and adults examined the efficacy of treatment after a first unprovoked seizure (6,20,28, 29, 30, 31). Two well-designed prospective studies which randomized subjects to treatment or placebo following a first unprovoked seizure found that treatment reduced the recurrence risk by approximately half (6,20,28). The larger Italian study included both children and adults (20,28). However, while recurrence risk was reduced, there was no difference in long-term outcomes between the two groups. An equal proportion were in 2-year remission after 5 years of follow up (28). Although the authors of this study initially recommended treatment following a first seizure, once it became apparent that early treatment did not affect long-term prognosis, they changed their recommendation, suggesting that in the majority of cases treatment wait until the second seizure (28). In general, the accumulating evidence from a large number of studies indicates that AED therapy is effective in reducing the risk of a recurrent seizure but does not alter the underlying disorder and therefore does not change long-term prognosis (32). Based on these data and assessment of risk-to-benefit, the American Academy of Neurology has issued a practice parameter on AED therapy following a first unprovoked seizure in children and adolescents (31). This parameter recommends that (a) treatment with an AEDs is not indicated for the prevention of the development of epilepsy, and (b) treatment with an AED may be considered in circumstances where the benefits of reducing the risk of a second seizure outweigh the risks of pharmacologic and psychosocial side effects. The authors rarely prescribe AEDs after a single seizure. A practice parameter addressing this issue in adults is currently under development.

What Happens After Two Seizures?

Two studies in adults (9) and children (14) examined what happens after a second seizure. In adults, the recurrence risk after a second seizure is 70%, leading Hauser and coworkers to conclude that, in adults, once a second seizure has occurred, treatment with AEDs is appropriate (9). In children, the recurrence risk following a second seizure is also approximately 70%. Those with a remote symptomatic etiology and those whose second seizure occurs within 6 months of the first have a higher recurrence risk (14). Interestingly, factors such as an abnormal EEG and sleep state at the time of the seizure, which help to differentiate those who only have one seizure from those who experienced a recurrence, are no longer associated with a differential risk of further seizures once a second seizure occurs (14). Despite the similarities in recurrence risk, the issue of treatment following a second seizure in children is less straightforward than in adults. Many of these children have idiopathic self-limited epilepsy syndromes, such as benign rolandic, where the need for treatment has been questioned (33, 34, 35). In addition, the frequency of seizures in this group is low, with only 25% of children who had 2 seizures experiencing 10 or more seizures over a 10-year period (14). Thus, the decision regarding treatment in children with cryptogenic/idiopathic seizures who have a second seizure must be individualized and take into account whether the seizures are part of a benign self-limited syndrome, as well as the frequency of the seizures and the relative risks and benefits of treatment.

WITHDRAWAL OF ANTIEPILEPTIC DRUGS IN THOSE WHO HAVE BEEN SEIZURE FREE ON ANTIEPILEPTIC DRUG THERAPY

AED therapy effectively controls seizures in the majority of patients with epilepsy. The preponderance of evidence indicates that most patients with epilepsy will become seizure free on AEDs within a few years of diagnosis (36, 37, 38, 39, 40, 41, 42, 43). However, the long-term use of AEDs carries with it significant morbidity. Therefore, the issue of whether one can withdraw AEDs in patients with epilepsy after a seizure-free interval becomes important in the treatment of a vast number of patients.

A large number of prospective and retrospective studies in children and adolescents, involving thousands of subjects, have been done over the past 25 years on the question of remission and relapse rates after withdrawal of AEDs. A smaller but still substantial number of studies dealing with adults have also been reported (37,44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76). A meta-analysis of the available literature reported a pooled risk of relapse of 25% at 1 year and 29% at 2 years following AED withdrawal (46).

In childhood-onset epilepsy, the majority of studies report that 60% to 75% of children and adolescents with epilepsy who have been seizure free for more than 2 to 4 years on medication will remain so after AEDs are withdrawn (44, 45, 46, 47,53, 54, 55,57,58, 59, 60, 61,63,68, 69, 70, 71,74,76). Exceptionally low recurrence rates of 8% to 12% were reported in studies that limited subject entry to neurologically normal children with normal EEGs, many of whom were followed since the onset of their seizures (62,73).

In the past, it was thought that adult-onset epilepsy had a far less favorable prognosis for remission than child-hood-onset epilepsy, and that withdrawal of medications was rarely feasible in this population. Although the prognosis in adults does appear to be worse than in children, newer studies suggest that the differences are smaller than was thought. Four years after onset, the majority of adults with new-onset seizures will be at least 2 years seizure free (41,42). Many adults self-discontinue their medications and are still seizure free years later (36,76). Studies of withdrawing AEDs in adults report recurrence rates of 28% to 66% (47,50,56,60,63,65,75), which is a much larger range than that reported in pediatric studies. However, it should be noted that studies that reported the lowest recurrence risks (50) limited themselves to patients followed since onset of their seizures and who had absence of other presumed risk factors. In pediatric studies, such selected populations have reported recurrence risks of less than 20%.

The preponderance of data at this time indicates that the recurrence risk following withdrawal of AEDs is somewhat higher in adult-onset epilepsy than in childhood-onset epilepsy with a relative risk of approximately 1.3 (46). However, much of the increased risk reported in some studies is a result of the higher risk of recurrence in adolescent-onset seizures (46,68). Selected populations of adults may have low recurrence risks. Two reports showed no differences in recurrence risks between children and adults (50,75). However, these studies have the highest reported recurrence risks for children (31% to 40%) and the lowest reported recurrence risks in adults (35% to 40%). In addition, their definition of children exceeds the usual limits of the term. In one study, 38% of the subjects had childhood onset but this was defined as onset before 15 years of age (50). Several studies in children have reported that an age of onset older than 10 or 12 years was associated with a higher recurrence risk, presumably because this already reflects early adult-onset epilepsy (48,67, 68, 69,73).

Only gold members can continue reading. Log In or Register to continue