Antiepileptic Drug Withdrawal in Patients with Temporal Lobe Epilepsy Undergoing Presurgical Video-EEG Monitoring


Address correspondence and reprint requests to Dr. C-H. Yiu at Department of Neurology, The Neurological Institute, Taipei Veterans General Hospital, 201 Sec.2, Shih-pai Road, Shih-pai, Taipei, Taiwan 11217 (ROC). E-mail:


Summary:  Purpose: To investigate antiepileptic drug (AED) withdrawal during video-EEG monitoring in adult patients with temporal lobe epilepsy (TLE).

Methods: Between 1995 and 1997, 102 consecutive patients with refractory TLE were admitted to the epilepsy monitoring unit for presurgical evaluation. Patients were monitored with ongoing AEDs being rapidly decreased and discontinued in 4–6 days. The monitoring was continued until sufficient numbers of seizures were recorded. Serum AED levels were checked at admission and after the first complex partial seizure (CPS).

Results: In all, 89 patients had 429 CPSs (mean, 4.8 per patient), including 156 (36.4%) secondarily generalized. A mean of 153.8 h (16–451 h) was required for completing the monitoring in each patient. Forty-three (48.3%) patients experienced seizure clusters, and eight (9.0%) had generalized seizures that had never occurred or had been absent for years. However, none evolved to status epilepticus. Carbamazepine was the most commonly used AED in 71.9% of patients, followed by valproate and phenytoin. When the first CPS occurred, mean 77.2 h since the beginning of the monitoring, serum levels of these three AEDs were mostly subtherapeutic rather than minimal.

Conclusions: Acute AED withdrawal effectively provoked seizures in TLE patients undergoing presurgical video-EEG monitoring. However, nearly 50% of patients had seizure clusters or secondarily generalized seizures. Serum AED levels were mostly subtherapeutic when the first CPS occurred.

Long-term video-EEG monitoring, allowing precise correlation of clinical events and associated EEG changes, is widely used in the evaluation of patients who were candidates for epilepsy surgery (1). However, because of the unpredictability of seizure paroxysms, it may require a lengthy stay in the epilepsy monitoring unit (EMU) to record sufficient numbers of seizures for decision making.

Discontinuation of antiepileptic drug (AED) treatment has been a routine to precipitate seizures in the EMU, in anticipation of reducing time and cost associated with the EEG monitoring. Several authorities have reported the effect of AED withdrawal in different groups of patients; however, rate and order of AED withdrawal varied greatly (2–4).

In this study, we rapidly decreased AEDs during video-EEG monitoring in adult Chinese patients with refractory temporal lobe epilepsy (TLE) undergoing presurgical evaluation.



This study encompassed 102 consecutive patients with TLE admitted to the adult EMU at Taipei Veterans General Hospital in Taiwan between August 1995 and August 1997 for presurgical evaluation. Their epilepsy appeared refractory to adequate doses of conventional and recently approved AED treatment for >2 years. In addition, all patients had evident epileptiform discharges in the temporal lobe(s) on serial scalp EEGs.

Video-EEG monitoring

Presurgical evaluation included several days of continuous video-EEG monitoring in the EMU using scalp electrodes and bilateral sphenoid electrodes. Intracranial EEG recording included foramen ovale electrodes and subdural grid electrodes that were implanted in seven patients. Patients were instructed to stay in bed preferably during the monitoring, and push the alarming button if there was a warning aura. At least a family member or friend should accompany the patient. An intravenous line with heparin lock was established in the forearm before the monitoring. All patients completed consent forms.

AED levels

We studied serum AED levels in all patients twice, at admission and after the first complex partial seizure (CPS). The following conventional AEDs were checked on the day of admission: carbamazepine (CBZ), valproate (VPA), phenytoin (PHT), phenobarbital (PB), and primidone (PRM). The second serum AED level check-up was done 15–20 min after the first CPS, along with prolactin determination for exclusion of nonepileptic seizures.

We arbitrarily classified serum AED levels into (a) therapeutic, including above or within therapeutic, (b) subtherapeutic, and (c) minimal (CBZ, <0.5 μg/mL; PHT, <0.5 μg/mL; VPA, <0.7 μg/mL; PRM, <0.3 μg/mL; and PB, <1.1 μg/mL).

In this study, therapeutic levels of AEDs in total form were as follows: CBZ, 4–12 μg/mL; PHT, 10–20 μg/mL; VPA, 50–100 μg/mL; PRM, 5–15 μg/mL; and PB, 15–35 μg/mL.

AED withdrawal

AEDs were reduced uniformly for all patients in this study. There was no AED withdrawal in the first monitoring day. On the second day, we completely discontinued (a) conventional AEDs with subtherapeutic levels or minimal levels; (b) other drugs including recently approved AEDs such as lamotrigine (LTG), vigabatrin (VGB), and clobazam (CLB). In other words, only conventional AEDs with therapeutic levels were kept after the second day. From the fourth day on, we decreased remaining AEDs simultaneously by about one third of their original daily doses in the following 3 days. So there were two-third original doses in the fourth day, one-third original doses in the fifth day, and no AEDs since the sixth day. Thereafter, patients were monitored with no AEDs for a maximum of 2 weeks.

Of patients taking only conventional AEDs at admission and all at therapeutic levels, AEDs were decreased simultaneously since the second monitoring day, and followed by the remaining procedures.


Video-EEG monitoring was continued by decision of the attending physicians until sufficient numbers of seizures with good visual and EEG quality were recorded. Then we resumed ongoing AEDs in a day. An ampoule of lorazepam (LZP, 2 mg) was given from the intravenous line if there were two consecutive secondarily generalized seizures in an hour or three CPSs in 24 h throughout the monitoring. In this study, we included for analysis only CPSs with or without secondary generalization that had the semiology and EEG characteristics of partial seizures of temporal onset set by the International Classification of Epilepsies and Epileptic Syndromes (5). In addition, all included CPSs were verified to be habitual ones by patients families or friends. Simple partial seizures or auras were not included. Patients who had no CPSs for a maximal 2-week stay after discontinuation of all AEDs at the EMU were also excluded.

The date and time of occurrence of each CPS in all patients were tabulated throughout the monitoring period. The time that was required for the first CPS to develop was expressed in hours with respect to the beginning of monitoring. The time that was required for completing a video-EEG monitoring was expressed in hours also, being the period from beginning of the monitoring to the last recorded CPS.


In all, we included 89 patients, 50 men and 39 women, aged 16–57 years (mean, 30.5 years), and seizure history ranged from 2 to 47 years (mean, 16.0 years). We excluded 13 patients who had only simple partial seizures, pseudoseizures, or no seizures during the monitoring.

Regarding the AEDs, 20 patients were receiving monotherapy, 44 patients were taking two AEDs, 21 patients were taking three AEDs, and four patients were taking four or more AEDs. A stable, daily basic regimen of benzodiazepines was considered as one of the AEDs. CBZ was the most commonly used AED in 64 (71.9%) patients, followed by VPA in 42 (47.2%) patients, PHT in 28 (31.5%) patients, CLB in 24 (27.0%) patients, LTG, PRM, PB, CZP, VGB, and acetazolamide in 11, eight, four, four, one, and one patients, respectively.

A total of 429 CPSs was collected in the 89 patients (mean, 4.8 per patient). Among these patients, 11 had two CPSs, 20 had three CPSs, 17 had four CPSs, 10 had five CPSs, 13 had six CPSs, and 18 had seven or more CPSs in the monitoring. Regarding CPSs, 156 (36.4%) evolved to secondary generalization.

The time required for completing a video-EEG monitoring for each patient ranged from 16 to 451 h (mean, 153.8 h). Fourteen (15.7%) patients completed the monitoring in three days, 45 (50.6%) patients in six days, and 44 (49.4%) patients needed ≥7 days. Figure 1 illustrates the length of stay in the EMU for all the patients.

Figure 1.

Length of stay for patients undergoing video-EEG monitoring in the epilepsy monitoring unit (EMU).

Forty-three (48.3%) patients had three or more CPSs within a 24-h period (seizure cluster), including 11 (12.4%) patients had more than five CPS. Fifty-one (57.3%) patients had at least one CPS that secondarily generalized. However, only eight (9.0%) patients who had never experienced generalized seizures or had not had them during the last 2 years experienced them during the monitoring. Status epilepticus did not occur in any of the patients.

Regarding specific AEDs, three (75%) of four patients receiving PB treatment and six (75%) of eight patients receiving PRM treatment had seizure clusters, including two with generalized seizures that had not previously occurred.

Changes of AED levels at admission and after the first CPS are shown in Table 1. Of patients taking CBZ, the majority (89.1%) achieved therapeutic levels on admission. In contrast, even with adequate doses of VPA and PHT treatment, only 16 (38.1%) of 42 patients and 12 (42.9%) of 28 patients had therapeutic levels, respectively.

Table 1.  Changes in serum AED levels for patients receiving polytherapy
AEDSerum AED levelsNA
At admission (no. of patients) (%)After the first CPS (no. of patients) (%)
  1. AED, antiepileptic drug; CPS, complex partial seizure; NA, not available; CBZ, carbamazepine; VPA, valproic acid; PHT, phenytoin; PB, phenobarbital; PRM, primidone.

CBZ57 (89.1)5 (7.8)2 (3.1)25 (39.1)27 (42.2)8 (12.5)4 (6.2)
VPA16 (38.1)26 (61.9)012 (28.6)20 (47.6)6 (14.3)4 (9.5)
PHT12 (42.9)16 (57.1)05 (17.8)18 (64.3)3 (10.7)2 (7.2)
PB2 (50)2 (50)004 (100)00
PRM7 (87.5)1 (12.5)004 (50)2 (25)2 (25)

The first CPS of each patient occurred 5–317 h (mean, 77.2 h) since the beginning of the monitoring. Figure 2 showed time distribution of the first CPS of each patient in the EMU. Fourteen (15.7%) patients had a CPS in the first monitoring day, 50 (56.2%) in 3 days, 64 (71.9%) in 4 days, 76 (85.4%) in 6 days, and only 13 (14.6%) needed >6 days.

Figure 2.

Occurrence of the first complex partial seizure of each patient in the EMU.

When the first CPS occurred, the AEDs were usually below therapeutic levels, more frequently at subtherapeutic than minimal levels (Table 1). Regarding patients with CBZ treatment, for example, 25 had therapeutic levels and 35 were below therapeutic levels including 27 at subtherapeutic and eight at minimal levels (CBZ levels were not available in four patients). Of patients with VPA and PHT treatment, most had subtherapeutic levels at admission and after the first CPS. In all, only eight (9.0%) patients had all remaining AEDs at minimal levels when the first CPS occurred.

Table 2 showed changes of AED levels in the 20 patients with monotherapy on admission. CBZ was the most common in 15 (75%) patients, with CBZ levels mostly therapeutic at admission and subtherapeutic when occurring after CPS, similar to that in patients with combined therapy. Of 12 patients taking CBZ monotherapy and with therapeutic levels, the first CPS occurred at 14–91 h (mean, 48.8 h), and 60–337 h (mean, 150.1 h) was required for monitoring.

Table 2.  Changes in serum AED levels for patients receiving monotherapy
AEDSerum AED levelsNA
At admission (no. of patients) (%)After the first CPS (no. of patients) (%)
  1. AED, antiepileptic drug; CPS, complex partial seizure; NA, not available; CBZ, carbamazepine; VPA, valproic acid; PHT, phenytoin.

CBZ12 (80)2 (13.3)1 (6.7)5 (33.3)7 (46.7)2 (13.3)1 (6.7)
VPA1 (100)00001 (100)0
PHT03 (100)002 (66.7)1 (33.3)0

Sixty-nine patients subsequently underwent temporal lobectomy (TL), 39 on the right and 30 on the left. Pathologic diagnosis of the resected specimens revealed hippocampal sclerosis in 61 patients, tumor lesions in seven patients, and arteriovenous malformation in one patient. Fifty-seven (82.6%) patients were seizure free for ≥2 years after TL (6).


In this study, we enrolled adult patients with refractory TLE. In addition, AED withdrawal was relatively rapid and steady in all patients. To prevent a peak activation of seizures during the monitoring, we first discontinued conventional AEDs that were below therapeutic levels and recently approved AEDs, which were often added on recently. A subsequent simultaneous reduction of remaining AEDs by reducing one third of original maintenance doses in 3 days was considered to be safe (2). Further, we analyzed CPSs only in this study because CPSs always mean a disruption of daily life to the patients because of impaired consciousness, and therefore could be a better predictor of surgical outcomes.

The mean length of stay in the EMU in our patients with TLE, 77.2 h (3.2 days) for the first CPS or 153.8 h (6.4 days) to end the monitoring, did not significantly differ from that of previous studies. Rak (7) found that 5.5 days were required to localize the seizure onset by noninvasive monitoring with AED withdrawal and sleep deprivation. However, seizure types of patients and methods of AED withdrawal were not mentioned. Using one-after-another AED discontinuation in 20 patients with TLE, Swick et al. (8) showed that 4.4 days were needed to record a partial seizure and 7.4 days for three habitual seizures, respectively. In 71 patients undergoing invasive EEG with varied AED withdrawal, Todorov et al. (9) concluded that 2.9–3.7 days were required for a seizure, 4.5–5.5 days for three seizures, and 6.1–7.6 days for five seizures. In this study, a mean of 4.8 CPSs was considered sufficient for each patient in the monitoring unit. Although 18 (20.2%) patients needed seven or more CPSs to better localize the seizure onset, about two thirds of patients needed three to six CPSs, and 11 (12.4%) patients with monitoring ended before the generally accepted three CPSs.

Short-term AED withdrawal was risky for provoking generalized seizures in nearly 50% of patients (2,10) and rarely led to complicated status epilepticus (11,12). With a more strict definition of three or more CPSs within 24 h than that of Mark et al. (3), seizure clusters occurred in 48.3% of our patients. High incidences of seizure clusters and generalized seizures (57.3%) in our patients could be attributed to rapid discontinuation of AEDs. Intravenous injection of LZP probably played a role in preventing status epilepticus, and might prolong the stay in the EMU. Whether oral, rectal, or intravenous infusion of other AEDs like DZP or PHT is better in this situation, as previously suggested (13–20), deserves further study. For the 75% of our patients with PB or PRM who experienced seizure clusters, we recommend a more cautious withdrawal of these two AEDs. William et al. (21) found that patients being withdrawn from PB had seizure frequency increase with serum PB <20 mg/L.

The first CPS of each patient was usually associated with below therapeutic serum AED levels, more at subtherapeutic than minimal, as noted in previous reports (2–4). Because most patients in this study were taking combination of two or more AEDs, the relationship of seizures to each AED was complicated because of possible drug interactions. This was well represented in that >50% of our patients receiving VPA or PHT had below therapeutic levels, even after adequate treatment doses. The relationship of AEDs and seizures was, nonetheless, more clearly represented in patients with CBZ monotherapy (Table 2), which demonstrated a tendency to occurrence of CPSs after decreasing serum CBZ levels. Regarding mechanisms of how AED discontinuation provokes seizures, either a loss of therapeutic effect of AEDs as shown in this study or a transient rebound phenomenon as occurring in cases of abstinence syndrome like alcohol or substance withdrawal may be responsible (10).

Acknowledgment: This study was supported in part by grants from Taipei Veterans General Hospital and Yen Tjing Ling Medical Foundation. We thank Dr. K.P. Lin for his invaluable assistance in data analysis.