Seizure Recurrence following a First Seizure

Studies assessing the prognosis for patients presenting with their first seizure or seizures have tended to focus on seizure recurrence rates. For patients presenting with a single seizure, many observational studies, both prospective and retrospective, have assessed seizure recurrence rates, many of which have been summarized in a systematic review and meta-analysis reported by Berg and Shinnar.¹ Thirteen first seizures studies were included in this systematic review, and the overall seizure recurrence risk was 46% (44 to 49) and for the subset of 5 prospective studies was 40% (37 to 43). The 2-year recurrence risk was 42% (39 to 44) across all studies and 36% (32 to 39) for the subset of prospective studies. There was a significantly higher recurrence risk for patients with partial onset seizures, an abnormal EEG, and neurological abnormality (learning difficulty, neurological signs, etc.). Thus, for the majority of patients presenting with a single seizure, the risk of a recurrence is low, and given the potential side effects of antiepileptic drugs, antiepileptic drug treatment has not been routinely recommended.

The most reliable method to assess the influence of antiepileptic drug treatment on recurrence rates and other outcomes is the randomized controlled trial. Such a study would randomly allocate patients to treatment or no treatment (or immediate or deferred treatment) and ideally should be large enough to allow subgroup analyses, most commonly in regression analyses, allowing the identification of patient characteristics that influence seizure recurrence risks and the magnitude of any treatment effect. To date, five published randomized controlled trials have examined the impact of antiepileptic drug treatment following a first seizure,^2–6 and the key characteristics of these trials are summarized in Table 17-1.

These trials consistently show a lower seizure recurrence rate with antiepileptic drug treatment compared to no treatment or placebo. There are, however, substantial differences among these trials in the estimated recurrence risks, with the 1- to 2-year recurrence risk ranging from 4 to 32% for AED-treated patients and 39 to 59% for untreated patients. These differences are most likely explained by differences in study design and patient populations in terms of factors such as time between seizure and randomization, seizure type, and age. Only two trials recruited sufficient patients to examine patient predictors of outcome. In the FIRST trial^4,7 age <16 years, secondarily generalized seizures, remote symptomatic seizures, and epileptiform changes on the EEG were associated with a higher risk of recurrence, whereas acute treatment with benzodiazepines was associated with a lower risk of recurrence.

The multicenter study of early epilepsy and single seizures (MESS trial)⁶ recruited 812 patients following a single seizure and 631 following two or more seizures. Regression modeling showed that the number of seizures before randomization, an abnormal EEG, and the presence of a neurological abnormality (learning disability or neurological signs) increased the risk of a seizure recurrence.⁸ This allowed the creation of a prognosis index to stratify patients in the groups with a low, medium, or high risk of seizure recurrence as summarized in Table 17-2. The recurrence risks estimated for these groups are given in Table 17-3.

TABLE 17–2 Prognostic Index from the MESS Trial

Reproduced with permission from: Kim LG, Johnson AL, Marson AG, Chadwick DW on behalf of the MRC MESS Study Group. Prediction of risk of seizure recurrence after a single seizure and early epilepsy: further results from the MESS trial. Lancet Neurol. 2006;5:217-222.

Seizure Recurrence following a Second Seizure

The risk of unprovoked seizures following a second seizure has not been examined in a prospective population-based study of patients untreated with antiepileptic drugs. Antiepileptic drugs became accepted practice for patients with epilepsy before robust epidemiological methods had been applied to examine the natural history of untreated epilepsy. It would now, of course, be considered unethical to deny a population of patients a treatment that is believed to be effective in order to examine this. The risk of seizure recurrence following two unprovoked seizures has been examined by Hauser et al.,⁹ who prospectively followed up 204 patients, 87% of whom were on antiepileptic drug treatment following their second seizure. The risk of a third seizure following a second was 32% (21 to 43) at 3 months, 41% (29 to 53) at 6 months, 57% (45 to 70) at 12 months, and 73% (59 to 87) at 4 years. The risk of a third seizure was higher in patients with remote symptomatic seizures than those with idiopathic or cryptogenic seizures, hazard ratio 1.9 (1.0 to 3.4). Other factors such as seizure type, EEG abnormality, or neurological abnormality were not significantly associated with a third seizure. The risk of a fourth seizure following a third was 31% (16 to 46) at 3 months, 48% (32 to 64) at 6 months, 61% (44 to 77) at 1 year and 76% (60 to 90) at 3 years. Shinnar et al. found similar results in a cohort of 407 children prospectively followed after their first seizure.¹⁰

Thus, in contrast to patients with a first seizure, the risk of an additional seizure following a second or third unprovoked seizure is high, even with the use of antiepileptic drugs. Given this finding in their paper, Hauser and colleagues recommend that patients with two or more unprovoked seizures should start antiepileptic drug treatment, although this study, by design, is unable to define the magnitude of any effect that antiepileptic drug treatment might have on seizure recurrence rates. To address this question, we need data from randomized controlled trials comparing treatment with no treatment for a cohort of patients with newly diagnosed epilepsy. No such trial has ever been undertaken and would now be considered unethical and would certainly contravene the Declaration of Helsinki.¹¹ At present, therefore, for patients with newly diagnosed epilepsy, we have no robust data that defines the magnitude of any treatment effect of antiepileptic drugs, both for the short-term outcomes (seizure recurrence) discussed earlier, as well as for the longer-term outcomes that are discussed later in this chapter. Despite this, consensus seems to hold that the majority of patients should be offered treatment following two or more seizures, particularly if they have occurred over a relatively short period of time (6 to 12 months). There has continued to be uncertainty regarding the appropriate recommendation for patients with seizures of minor symptomatology (e.g., simple partial seizures), and patients with long periods of time between seizures. Such patients (631) were entered into the MESS study⁶ when both clinician and patient were uncertain about the need for antiepileptic drug treatment.

Longer-Term Outcomes

The main focus of antiepileptic drug treatment is to control seizures, thus increasing the probability of a long-term remission from seizures, primarily via the antiseizure effects of antiepileptic drugs. In contrast to the risk of seizure recurrence following a first or second unprovoked seizure, the impact of immediate versus deferred treatment on longer-term outcomes, including remission rates, has been less well studied. Whereas preventing a seizure recurrence might represent an “antiseizure” effect of antiepileptic drugs, there has been ongoing debate as to whether antiepileptic drugs might have an “antiepileptogenic” effect and alter the natural history of epilepsy.

Given that up to 30% of patients with epilepsy are drug refractory¹² and experience continued seizures and a diminished quality of life, any intervention proven to improve the natural history, by reducing the severity of epilepsy and improving the probability of a long-term remission from seizures, could have important consequences for a large number of patients.^13,14 Thus it is important to understand whether the early use of antiepileptic drugs might influence the natural history of epilepsy.

Some have suggested that epilepsy might be a self-facilitating process in which seizures beget seizures,^15,16 whereby each seizure causes further neuronal damage and reorganization that increases the probability of the next seizure. This idea was suggested by Gower in 1881, when he suggested that “the tendency of the disease is toward self perpetuation; each attack facilitating the occurrence of the next by increasing the instability of nerve elements.”¹⁷

Elwes et al.¹⁸ recruited a cohort of 183 patients presenting with a new diagnosis of epilepsy and retrospectively assessed the time interval between successive tonic-clonic seizures prior to their clinic visit. They observed that the time interval between successive seizures became progressively shorter, and the authors suggest that this finding supports early epilepsy being an accelerating process. However, this study may be confounded by its retrospective design, selection bias, seizure type, and perhaps seizure clustering and does not provide evidence of seizures begetting seizures.

Other studies have showed that patients with more seizures or a higher seizure frequency prior to starting antiepileptic drug treatment had a lower rate of remission from seizures.^19,20 This finding was again interpreted as indicating that seizures predispose to further seizures and used as an argument for early treatment. However, the number of seizures prior to starting treatment is likely to be correlated with seizure type, with patients with complex partial seizures experiencing a greater number of seizures before starting treatment and having a poorer prognosis, as was the case in these studies. Thus although seizure type and the number of seizures prior to starting treatment may be important prognostic factors, these studies do not necessarily provide evidence of seizures begetting seizures.

Two studies have been undertaken in developing countries, one from Kenya²¹ and one from Ecuador,²² in which people with untreated epilepsy were ascertained by field workers going door to door. These patients were then offered entry into a randomized controlled trial comparing carbamazepine and phenobarbital; 192 entered the trial in Ecuador, and 302 entered the trial in Kenya. Despite having untreated epilepsy, sometimes for many years, in both studies 53% of patients were seizure free from the 6th to the 12th month of follow-up, which is comparable to outcomes in randomized controlled trials in newly diagnosed populations.^23,24 This finding would suggest that the failure to treat epilepsy early does not worsen the prognosis of epilepsy once treatment is started.

There is some animal data that suggests that some antiepileptic drugs may exhibit antiepileptogenic properties. Studies in kindled rats suggest that valproate, phenobarbital, diazepam, vigabatrin, topiramate, levetiraceta, and lamotrigine might have antiepileptogenic properties,²⁵ although the appropriateness of the kindling model for assessing antiepileptogenic properties has been criticized.²⁶ Post-status epilepticus models suggest that topiramate and valproate may have antiepileptogenic properties, although results have been conflicting. Although these models may be appropriate to examine potential antiepileptogenic effects and might equate to patients presenting with prolonged status epilepticus, they do not readily equate to the patient presenting with a single seizure or early epilepsy.

Assessing the Impact of Antiepileptic Drug Treatment on Longer-Term Outcomes

Reliable evidence regarding the impact of antiepileptic drug treatment on longer-term outcomes for patients presenting with a single seizure or early epilepsy can only be gained from randomized controlled trials. Such trials will recruit patients for whom there is uncertainty about the need for immediate AED treatment, primarily those with a single seizure, as well as those with two or more seizures if the seizures are of minor symptomatology (e.g., simple partial seizures) or separated by significant periods of time. Patients will be randomized to immediate treatment with an antiepileptic drug or to deferred treatment. Patients randomized to deferred treatment should be allowed to start treatment once the clinician and patients agree that this is required, primarily following further seizures. Thus such trials do not compare treatment with no treatment, but compare immediate with deferred treatment and hence assess the impact of preventing a small number of seizures with immediate treatment on outcome. Given that less than 50% of patients with a single seizure will have a recurrence in the first 1 to 2 years, the policy of immediate treatment does expose a large number of patients that would not have had a seizure recurrence to antiepileptic drug treatment, the side effects and consequences of which should be assessed.

Although double-blind trials are often considered a gold standard, particularly for short-term drug regulatory trials, blinding is not always practical or desirable in trials designed to inform clinical decision making when comparing differing treatment policies, particularly when long-term follow-up is required. Any assessment of the effect of early treatment on longer-term outcomes requires patients to be followed up for a number of years, which along with the need to initiate treatment in patients allocated to deferred treatment who have a seizure recurrence, makes blinding impracticable in this scenario.

The longer-term seizure outcome most commonly examined is seizure remission rates, which is one of the primary outcomes recommended by the International League Against Epilepsy for trials of antiepileptic drug monotherapy.²⁷ One approach is to use survival methods to assess the time taken from randomization to achieve a specified period of remission (e.g., 1 or 2 years). This approach allows an assessment as to which treatment policy is associated with achieving this outcome soonest, most commonly expressed as a hazard ratio or risk ratio. In addition, this approach allows an estimate of the proportion of patients that have achieved a remission at specific points in time.

A second approach is to examine what has been called terminal remission rates. Rather than examine remission rates from randomization, this approach examines the proportion of patients seizure free in a certain time window, such as between the third and fifth year after randomization, referred to as a 2-year terminal remission at 5 years. The natural history for some patients is to enter an early remission and then have a recurrence, some of whom will develop a refractory epilepsy. Such patients would be categorized as achieving a remission in an analysis of time to achieve a remission from randomization, but would be categorized as not in remission in an assessment of terminal remission rates, provided the time window is appropriate. Thus an assessment of both time to remission from randomization and terminal remission rates are important outcomes to assess when examining the influence of early antiepileptic drug treatment on the natural history of epilepsy. Given that both antiepileptic drug treatment and seizures are associated with adverse events and changes in quality of life, it is important that these outcomes are also assessed in any trial comparing immediate and deferred treatment for patients with single seizures and early epilepsy.

Three reports assessing longer-term outcomes in randomized controlled trials have been published.^28–31 Camfield et al.² reported a trial that recruited 31 children within 1 month of a first seizure who were allocated to immediate treatment with carbamazepine or deferred treatment. In 2001 they attempted to collect 15-year follow-up data²⁸ and were able to trace 26 of the original 31 participants, 16 of the 17 control group patients, and 10 of the 14 patients allocated carbamazepine. Four patients allocated carbamazepine, and five patients in the control group had remained free of seizures since randomization. Of the 10 patients originally allocated carbamazepine, eight had a 2-year terminal remission at 15-year follow-up, compared to 14 of the 16 allocated to the control group, RR 0.73 (0.24 to 2.2). Thus the number of patients with either a recurrence or in a 2-year terminal remission at 15-year follow-up was similar. The authors concluded that the long-term clinical course of epilepsy is not improved by treatment following a first unprovoked seizure. However, with 31 patients, this study did not have the power to exclude the possibility of an important treatment effect, as supported by the confidence intervals around the RR for a 2-year terminal remission at 15-year follow-up, 0.73 (0.24 to 2.22), which indicates that these results could be in keeping with doubling this terminal remission rate with early antiepileptic drug treatment.

For the FIRST trial, two reports give longer-term follow-up data, the first of which provides results for the first 7 years of follow-up. Eighty-seven percent of patients allocated immediate treatment and 83% allocated deferred treatment achieved a 1-year remission from seizures during follow-up. The estimated hazard ratio for time to achieve a 1-year remission was 1.2 (0.95 to 1.37), indicating that patients allocated immediate treatment may be 20% more likely to enter a 1-year remission, although this result was not statistically significant as the confidence interval crosses one.

They estimate that 63% in the deferred treatment group and 78% in the immediate treatment group entered a 1-year remission within the first 3 months of follow-up. At 2 years, it was estimated that 82% in the deferred treatment group and 85% in the immediate treatment achieved a 1-year remission, whereas at 8 years follow-up, 92% in the deferred treatment group and 93% in the immediate treatment achieved a 1-year remission. These results indicate that although patients allocated immediate treatment will enter remission sooner, by 3 or more years of follow-up there is little difference between the groups for this outcome. However, the authors do not give an estimate with confidence intervals for a difference between randomized groups.

The latest report from the FIRST trial provides data following 10 to 13 years of follow-up and provides results for 2-year remission rates.³⁰ For the whole cohort, 87% of those allocated immediate treatment and 78% of those allocated deferred treatment had achieved a 2-year remission. When the analysis was restricted to patients with 9 or more years follow-up data available, 92% of patients allocated immediate treatment and 93% allocated deferred treatment achieved a 2-year remission. The report estimates that during the first 3 months of follow-up 57% allocated deferred treatment and 72% allocated immediate treatment entered a 2-year remission. At 5 years follow-up, the report estimated that 79% allocated deferred treatment and 84% allocated immediate treatment had achieved a 2-year remission, whereas at 12 years of follow-up, 86% allocated deferred treatment and 85% allocated immediate treatment achieved a 2-year remission. Again, these results indicate that patients allocated immediate treatment might enter a 2-year remission sooner, but by 12 years of follow-up, the estimates are similar for both groups. Estimates for the differences between groups at these time points with 95% confidence intervals are not provided; hence, from the published data, we cannot be certain whether the possibility of important differences existing has been excluded.

The largest trial to examine the influence of early antiepileptic drug treatment on longer-term outcomes following a single seizure or early epilepsy is the MESS trial.⁶ This trial recruited 1443 patients, 812 following a single seizure and 631 following two or more seizures, the latter representing a group with infrequent seizures or seizures with minor symptomatology for whom there was uncertainty about the need for antiepileptic drug treatment.

The results for time to a 2-year remission are summarized in Table 17-4 and Figure 17-1. Results for the whole cohort recruited to the MESS trial show that at 2 years, 64% of the immediate-treatment group and 52% of the deferred-treatment group achieved an immediate 2-year remission, whereas at 5 years 92% of patients allocated immediate treatment had achieved a 2-year remission compared to 90% in the deferred-treatment group. The difference between the two groups at 5 years was 2% (–1.2 to 6.1). The confidence intervals around this estimate are narrow and exclude the possibility of an important effect of treatment. At 8 years, 95% allocated immediate treatment had achieved a 2-year remission compared to 96% in the deferred-treatment group. The difference between the two groups was 1% (–2.5 to 3.9), which again excludes the possibility of an important treatment effect. Thus, in keeping with results from the FIRST trial group, MESS confirms that patients starting treatment following a single or few seizures will enter a 2-year remission sooner than those allocated deferred treatment, but treatment has no influence on the proportion of patients entering a 2-year period of remission after 5 years or longer of follow-up.

Figure 17–1 Plots showing cumulative proportion of patients with first seizure after randomization, by treatment group and stratified by number of seizures reported at randomization in the MESS study.

Reproduced with permission from Marson A. Jacoby A. Johnson A. Kim L. Gamble C. Chadwick D, Medical Research Council MESS Study Group. Immediate versus deferred antiepileptic drug treatment for early epilepsy and single seizures: a randomised controlled trial. Lancet. 2005;365(9476):2007-2013.

The MESS trial also examined the proportion of patients in a 2-year terminal remission at 3 and at 5 years (i.e., the proportion of patients seizure free between years 1 and 3, and between years 3 and 5 of follow-up). At 3 years, 74% of the immediate-treatment group and 71% of the deferred-treatment group were seizure free between years 1 and 3 of follow-up, and the difference between the two groups was estimated as 3.4% (95% CI –1.6%, 8.5%). At 5 years, 76% of the immediate-treatment and 77% of the deferred-treatment group were seizure free between years 3 and 5 of follow-up and the difference between the two groups was estimated as –0.2% (95% CI –5.8%, 5.5%). Thus by year 5 of follow-up, there was no difference between the randomized groups for this outcome, and the confidence intervals around this estimate are narrow, excluding the possibility of an important treatment effect. These results also indicate that early treatment following a single or few seizures has no influence on longer-term seizure control in epilepsy.

Do the Findings of the FIRST and MESS Studies Apply to All Antiepileptic Drugs?

Both the FIRST and the MESS studies compared the policies of immediate and deferred treatment but allowed the investigators to choose which drug to start. In the FIRST study, of those randomized to immediate treatment, 47% received phenobarbital, 30% carbamazepine and 16% valproate. In the MESS study, 46% received carbamazepine, and 46% valproate. We therefore have strong evidence that neither immediate treatment with the first-line antiepileptic drugs carbamazepine and valproate nor with phenobarbital influences longer-term seizure outcomes in epilepsy compared to deferred treatment. Given our lack of understanding of the process of epileptogenesis and the mechanisms that might influence it, it is not possible to state with any confidence whether any of the other antiepileptic drugs, particularly the newer generation of drugs, might have any influence on the natural history of epilepsy. Any trial designed to find out whether they do would need to be of similar design and magnitude to the MESS study.

Adverse Effects and Quality-of-Life Consequences of Immediate versus Deferred Treatment

The MESS study reported data for adverse effects and for quality-of-life outcomes. Patients in the immediate-treatment group were more likely to report at least on adverse effect—39% versus 31%, difference 8% (4 to 14). The adverse events were largely those potentially associated with antiepileptic drug treatment. There were 54 deaths, 31 in the immediate-treatment group and 23 in the deferred-treatment group, 6 of which were classified as sudden unexpected death in epilepsy, four in the immediate-and two in the deferred-treatment group. Of the 527 eligible adults recruited into MESS in the U.K. 441 provided quality-of-life data at baseline and at 2 years.³² Overall there was no quality-of-life advantage for either treatment policy, with results suggesting a trade-off between the adverse consequences of more seizure recurrences in the deferred-treatment group and a higher rate of antiepileptic drug adverse events in the immediate-treatment group.

Conclusions

We have good evidence that for patients with a single seizure or early epilepsy, when compared to deferred treatment, immediate treatment with one of the current first-line antiepileptic drugs reduces the risk of seizure recurrence, but has no impact on the longer-term seizure outcome in epilepsy. Consensus holds that antiepileptic drug treatment should be initiated for the majority of patients who have had two or more seizures, as the risk of recurrence is high, although the aim of treatment is to control seizures rather than influence the natural history of epilepsy. For patients with a single seizure, the recurrence risk is generally low, and there is a trade-off between preventing seizures and adverse effects of antiepileptic drugs such that neither policy is associated with overall quality-of-life gains. Patients at a low, medium, and high risk of seizure recurrence can be identified to aid patients and clinicians in discussions regarding the pros and cons of starting antiepileptic drug treatment following a single seizure. Whether any of the newer generation of antiepileptic drugs might influence the natural history of epilepsy is yet to be investigated in human studies.