Patients with epilepsy often do not have spontaneous interictal epileptiform discharges (IEDs) on their routine EEG recordings. Such discharges are not present on the initial EEG recordings of about 40% of patients with newly diagnosed epilepsy.⁵² Approximately 10% of candidates for epilepsy surgery do not have IEDs, even with long-term monitoring.⁷ The possibility of suppression of IEDs by AEDs has often been raised.⁸⁰ Valproate consistently suppresses generalized spike-wave discharges.²⁹ Long-term treatment with carbamazepine seems to be associated with less frequent epileptiform discharges.⁶⁰ Intravenous phenytoin also can suppress generalized IEDs, but the effect is only transient, lasting for 10 to 20 minutes.⁵⁴ A single intramuscular dose of clonazepam has been shown in children to decrease significantly both generalized and focal epileptiform discharges.⁸ Selective suppression of some types of epileptiform discharges by AEDs has also been observed. Lamotrigine selectively suppresses long episodes of epileptiform discharges, but not short episodes of discharges.¹⁶ The effect of AEDs on cortical excitability has been studied recently in healthy volunteers by using transcranial magnetic stimulation.³⁹ The threshold for activating the motor cortex was increased with increasing total and free serum concentrations of carbamazepine and lamotrigine, although acute drug withdrawal decreased the threshold in only a small proportion of patients.

Barbiturate withdrawal in addicted persons who were nonepileptic reportedly has been followed by paroxysmal EEG activities.⁸⁶ However, according to currently accepted terminology, many of those reported EEG activities are nonepileptiform. Ludwig and Ajmone Marsan⁴⁴ observed that spikes developed after abrupt drug withdrawal in 10 of 13 patients who had had no EEG discharges while taking AEDs. A discrete epileptiform focus developed after drug withdrawal in 57% of all their patients. However, Gotman and Marciani²⁷ found no correlation between serum AED concentrations and spiking rate after tapering the dose. They showed that spiking activity increased after, and not before, seizure occurrence, and that the increase could last several days. Their findings raised the possibility that increased spike discharges previously attributed to drug withdrawal may have actually been a postictal phenomenon. Their method of computerized detection of spike discharges allowed them to quantitate IEDs continuously during several days. The application of this technology to a kindling model of epilepsy in cats yielded similar findings. Moreover, an earlier study showed that carbamazepine withdrawal in primate models of partial epilepsy failed to intensify spike discharges.¹⁰ Epileptiform discharges that occur after seizure episodes may be more widespread in distribution or contralateral to the focus of seizure onset.²⁶ Localization of the surgical focus must not rely solely on these discharges.

Numerous studies have investigated the effect of discontinuing AEDs on seizure characteristics. The common experience among investigators is that fairly abrupt discontinuation of an AED does increase seizure frequency. However, disagreements exist about the factors most responsible for the increase. Some investigators observed that increases occurred early as AED dosage⁵⁰ or concentration declined, rather than when the dose became minimal,⁸⁷ and the exacerbation of seizure frequency
was self-limited. It did not always persist toward the end of the tapering process. Hence, the increase in seizures has been attributed to a transient “rebound phenomenon” similar to that seen in abstinence syndromes of drug addiction,^18,32 rather than to a loss of therapeutic effect. In a study of patients withdrawing from phenobarbital, Theodore et al.⁷⁵ also observed that seizure rates were highest when serum concentrations passed below 15 μg/mL, but not when concentrations were very low.

Other studies^5,51 have concluded that loss of therapeutic effect, and not rebound phenomenon, is the basis for increased seizures during AED withdrawal. They found that seizure frequency increased when serum drug concentrations became subtherapeutic and not as they were falling from their baseline values. However, it is difficult to compare these studies critically for the following reasons: (a) drug regimens were not the same between studies, (b) the order or rate of withdrawal was not uniform, (c) serum drug concentrations were not determined using identical schedules, and (d) definitions of withdrawal and baseline periods were either not reported or were dissimilar. Furthermore, the contribution of falling serum levels to seizure exacerbation cannot be completely discounted, because the observation periods in patients undergoing video-EEG monitoring are typically limited.

Despite the difficulty in attributing seizure exacerbation during withdrawal to rebound phenomenon, abrupt discontinuation of barbiturates¹⁸ and benzodiazepines³² is unequivocally known to produce an abstinence syndrome, with withdrawal seizures occurring even in nonepileptic persons. The seizures develop only within the duration of the abstinence syndrome. They almost always are generalized convulsive episodes without focal onset. Animal studies suggest that these seizures are subcortical in origin.¹⁷ Hence, there is a theoretical possibility of nonlocalizing rebound seizures confounding the identification of the onset of seizures recorded during long-term monitoring. However, rebound seizures have been reported only in a setting of acute withdrawal from prolonged exposure to very high doses of barbiturates and benzodiazepines. Most cases involved short-acting barbiturates (e.g., secobarbital and pentobarbital) that are not used for chronic treatment of epilepsy. In patients with epilepsy who are taking clinically appropriate doses, the long half-life of phenobarbital probably prevents acute abstinence syndrome and rebound seizures from developing. In fact, seizures may not increase until several weeks after phenobarbital is discontinued,⁷⁵ whereas rebound seizures in persons addicted to barbiturates occur a few days after medication withdrawal.¹⁸

Concern has been raised about the possibility that AED withdrawal may precipitate seizures that are new and not typical of habitual or baseline seizures. However, the clinical and EEG onset characteristics of seizures following AED withdrawal are not substantially different from those of baseline seizures.⁸⁸ AED withdrawal does not affect seizure onset, although it does increase seizure frequency and duration and the chance of secondary generalization. Therefore, seizures precipitated by AED withdrawal are generally reliable for localization for epilepsy surgery, providing that their onset characteristics are verified to be identical to those of habitual seizures. Nonetheless, previously unrecognized seizure foci may appear after AED withdrawal. Spencer et al.⁷³ evaluated the video and depth electrode recordings of 71 baseline and 89 withdrawal seizures in 18 patients. During AED withdrawal, four patients had seizures with EEG or clinical features that were not typical of habitual seizures. One of these four patients had a poor surgical outcome despite the fact that the habitual seizures during baseline originated from a single focus. The authors warned that new seizure foci unmasked by drug withdrawal may be sources of recurrent seizures after surgery. In contrast, Engel and Crandall¹⁴ reported a patient who did well after right temporal lobectomy despite clinically different seizures developing from the contralateral side during AED withdrawal.

The potential for withdrawal seizures appearing to be falsely localizing was assessed further by Marciani and Gotman⁴⁹ in 14 surgical candidates. The majority had bilateral independent foci on baseline scalp and sphenoidal EEG. Only one patient had a seizure that was unlike habitual seizures, and it arose from an independent focus. This particular seizure may not have resulted from AED withdrawal, because it occurred before the AED dose was substantially reduced. This observation was corroborated later by Marks et al.,⁵¹ who found that most of the notably different seizures in their patients actually occurred outside the withdrawal period.

Nearly half of all patients undergoing AED withdrawal during long-term monitoring experience generalized convulsive seizures.⁵⁰ These seizures begin clinically and electrographically as partial seizures that are no different from the patients’ habitual seizures.⁴⁹ The duration of either partial or secondarily generalized seizures is usually unaffected by AED dose reduction.^70,74 Antiepileptic drug withdrawal does not alter the pattern of ictal discharges at the onset of these seizures, nor does it hasten spread to the contralateral hemisphere.

The preceding evidence indicates that, with rare exceptions, seizures recorded during AED withdrawal are reliable for localization of the ictal focus. Nonetheless, regardless of the status of AED medication and the number of foci detected, recorded seizures should always be verified with the patient and with witnesses to ensure that they are the same as the patient’s habitual and disabling seizures. If there is evidence of more than one ictal focus, further inquiry should be made to determine which seizures are primarily responsible for the patient’s disability. Constant vigilance is needed so that occult seizure foci are not overlooked. In patients with multiple seizure foci, clusters of seizures precipitated by drug withdrawal may arise from only one focus. Haut et al.³⁰ initially reported that in patients with bilateral seizure foci, seizures that occur within 8 hours of each other are more likely to arise from the same side than seizures that occur more than 8 hours apart. However, their subsequent study demonstrated that mean interseizure intervals were not significantly different for concordant and discordant seizures.³¹ The discovery of multiple ictal foci does not preclude surgery if it can be established that habitual seizures originate from one surgically resectable focus.^14,71 The identification of the predominant focus does not appear to be affected by seizure clustering.³¹

The profile of AED withdrawal seizures derived from the literature is based mainly on collective experience with intractable partial epilepsy. Therefore, the information obtained may not be applicable to patients with primary generalized seizures or other specific epilepsy syndromes. The effect of AED withdrawal on seizure types other than partial seizure disorders needs to be studied.

Withdrawal of AEDs can result in major complications, some of which are life threatening. Of patients undergoing AED withdrawal, about half experience generalized convulsions—many of whom have never had convulsive seizures before withdrawal. Furthermore, seizures exacerbated by AED withdrawal may lead to seizure clustering in 50% of patients^19,66,87 and to status epilepticus in 3%.⁶¹ The risk factors for seizure clustering are history of seizure clustering at home, mesial temporal sclerosis documented with magnetic resonance imaging, and more than one seizure onset zone.³¹ The rate of AED withdrawal and the location of the EEG ictal onset are not risk factors for seizure clustering in the epilepsy monitoring unit.

It should be noted that even isolated seizure episodes can result in falls that cause fractures, joint dislocations, and external or internal soft tissue injuries. The elderly are especially susceptible to compression fractures of the vertebrae.⁴⁷ They
also may be more vulnerable to the cardiac and respiratory effects of convulsive seizures.

Many of the potential complications of AED withdrawal can be anticipated and prevented if patients are under continuous observation by experienced staff. Patient areas should be designed to minimize the risk for injury from falls or unrestrained postictal behavior. Patients should be accompanied when walking. They should be attended to after every seizure. The time of occurrence and the intensity of each seizure should be noted. When seizures are unacceptably prolonged or increasingly frequent, AED treatment should be considered. Theodore et al.⁷⁴ observed that generalized tonic–clonic seizures that stop spontaneously do so within 2 minutes. They advocate the initiation of intravenous AEDs when generalized tonic–clonic seizures last longer than 2 minutes. As a general rule, the authors of this chapter administer an oral dose of AED when two or more generalized tonic–clonic seizures occur within 12 hours. When they occur within 2 hours, either an oral^78,79 or intravenous AED is given at a dose sufficient to achieve the desired serum concentration as promptly and as safely as possible. (As part of the protocol, intravenous access is established and maintained with a heparin lock when the patient is admitted.) The AED used in most instances is preferentially a drug that is being withdrawn. However, if intravenous administration is indicated, phenytoin is generally used, unless immediate cessation of prolonged seizures requires intravenous benzodiazepines. For the treatment of seizure clusters, sublingual lor-azepam in adults⁴⁷ or rectal diazepam in children⁴² may be used.

To manage life-threatening complications of AED withdrawal, medical staff certified in cardiorespiratory resuscitation must be readily available. The monitoring laboratory must be equipped with instruments and medications for resuscitation. Benzodiazepines and short-acting barbiturates should be withdrawn cautiously, particularly if they are taken at a high dose because of the development of tolerance. Seizures and other symptoms of withdrawal may be particularly severe.^18,72 A state of agitation and dysphoria can follow carbamazepine withdrawal.¹¹ In our experience, temporary resumption of carbamazepine or the use of an oral benzodiazepine alleviates the symptoms.