Increased Seizure Frequency Associated with Felbamate Withdrawal in Adults

ORIGINAL CONTRIBUTION Increased Seizure Frequency Associated With Felbamate Withdrawal in Adults

Timothy E. Welty, PharmD; Michael Privitera, MD; Rakesh Shukla, PhD

Objective: To characterize changes in seizure fre- patients who received felbamate were evaluated for quency following felbamate withdrawal. adherence to inclusion and exclusion criteria. Statistical analysis was performed using a log-linear model for Design: Nonrandomized, retrospective chart review of count data. a case series. Main Outcome Measures: Seizure frequency during Setting: Epilepsy program specializing in adults with the 6 months before initiating felbamate therapy served uncontrolled epilepsy. as the baseline. Changes in seizure frequency were evaluated by comparing the number of seizures in the Patient Population: Forty-five ambulatory patients month felbamate was tapered and the 3 months after withdrawn from felbamate use. Patients were included felbamate discontinuation with the baseline frequency. if they had received felbamate for at least 1 month, were Comparisons were made between patients who started 18 years or older, had accurate seizure frequency docu- gabapentin therapy and those who did not and between mentation, had accurate documentation of all antiepi- felbamate responders and nonresponders. leptic drugs, and received the same concomitant antiepileptic drugs before and after felbamate therapy, except Results: Felbamate withdrawal resulted in a signifi- for the possible addition of gabapentin. Patients were ex- cant (P=.02) increase in seizure frequency. Patients re- cluded if they had hematologic or hepatic toxic effects ceiving gabapentin had a smaller increase in seizure fre- with felbamate, were unable to withdraw from fel- quency, but the difference was not statistically significant. bamate treatment, had a progressive neurologic disor- There was no statistically significant difference in sei- der, or participated in another drug trial. zure frequency between felbamate responders and nonresponders. Methods: When information became available on aplastic anemia and hepatotoxicity associated with fel- Conclusions: Felbamate withdrawal caused a signifi- bamate, all patients were advised to taper their felba- cant increase in seizure frequency over the subsequent mate dosage over approximately 2 weeks. They received 3 months. These findings are important for clinical trial written instructions for tapering felbamate and adjust- design and clinical practice. ing concomitant antiepileptic drugs and kept calendars to note the number of seizures. The charts of all Arch Neurol. 1998;55:641-645

ITHDRAWING an an- who converted from AED polytherapy to tiepileptic drug monotherapy had an increased seizure fre- (AED), either to quency. Bromfield and colleagues3 pro- discontinue medi- vided additional support for this mecha- cation or to change nism when they showed that seizures toW another medication, may lead to in- increased only when phenytoin serum con- creased seizure frequency. At least 3 centrations become subtherapeutic or un- mechanisms have been proposed for this detectable. Finally, the removal of a drug- From the College of Pharmacy phenomenon. An AED may cause a true drug interaction may result in increased (Dr Welty), Department of withdrawal effect, producing increased sei- seizures. For example, when an AED that Neurology (Dr Privitera), and zures while the AED serum concentra- inhibits the metabolism of another AED the Division of Epidemiology tion is decreasing. This has been ob- is discontinued, an unwanted decrease in and Biostatistics, Department 1 of Environmental Health served during barbiturate withdrawal. the serum concentrations of the remain- (Dr Shukla), University of Increased seizures may also occur being AEDs will result and breakthrough sei- Cincinnati Medical Center, cause of a lack of therapeutic effect. zures will occur. In many cases, more than Cincinnati, Ohio. Schmidt2 reported that 17% of patients 1 mechanism may be responsible.

ARCH NEUROL / VOL 55, MAY 1998 641

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 PATIENTS AND METHODS Seizure frequency (all seizure types combined) during the 6 months before starting felbamate therapy (prefelbamate) was compared with the frequency during the month All patients treated with felbamate in our clinic were iden- felbamate was tapered and 3 months after felbamate dis- tified for review. This outpatient practice consists prima- continuation (postfelbamate). In this type of longitudinal rily of adults referred for treatment of epilepsy uncon- study design, repeated observations of seizure frequency trolled by standard AED therapy. When we were informed in individual patients enable a direct study of change in fre- of the Food and Drug Administration action, we in- quency. However, the set of observations in an individual structed our patients to taper the felbamate dosage over ap- patient may be interdependent, requiring special statisti- proximately 2 weeks. Each individual was given a written cal methods that account for this correlation before draw- schedule for tapering felbamate and adjusting concomi- ing valid scientific inferences. We chose the GENMOD pro- tant AED therapy. Seizure frequency from patient calen- cedure with repeated statements using SAS/STAT software dars had been recorded in the charts for each month re- (release 6.12, SAS Institute, Cary, NC) as a statistical method viewed. All charts were reviewed by a research nurse, who that considers this type of correlation. This statistical analy- is familiar with AED studies; a clinical pharmacist (T.E.W.), sis is based on using each patient’s seizure history indi- who specializes in AED pharmacotherapy; or an epilep- vidually. It provides the optimal solution to model param- tologist (M.P.). eters. Details of this approach can be reviewed elsewhere.5 Patients were included if they (1) had received fel- A log-linear model with ␯(µ)=µ (the Poisson variance func- bamate for at least 1 month; (2) were at least 18 years old; tion) and the following formula was used: (3) had accurate documentation by seizure calendars of to- log[E(Yij)] = β +Xiβi +Xiβ +XiXiβ + log(tij), tal monthly seizure frequency for 6 months before fel- 0 1 1 2 2 1 2 3 bamate use, during felbamate use, and at least 4 months whereYij indicates the number of seizures in interval j; tij after felbamate discontinuation; (4) had documentation in indicates the length of interval j (prefelbamate j=6, post- the chart of all AED doses and the rate of dosage change felbamate j=1) (to be used in log[E(Yij)]/tij); Xi1 is 0 prefel- (including any intermittent use of benzodiazepines), both bamate and 1 postfelbamate; and Xi2 is 0 for no gabapen- before and after felbamate discontinuation; and (5) re- tin, female sex, and felbamate nonresponders and 1 for ceived the same AED at similar doses during the first 4 gabapentin, male sex, and felbamate responders (patients months after felbamate was discontinued as during the 6 with a 50% or greater reduction in seizure frequency while months before felbamate was started, except for the pos- receiving felbamate). sible addition of gabapentin. For example, the difference between the log seizure We excluded patients from this study if they had rates in the baseline period and the posttreatment period β β β (1) hepatotoxic effects or severe hematologic adverse ef- is 1 for the no-gabapentin group and 1+ 3 for the gaba- β fects during felbamate treatment; (2) an inability to suc- pentin group. A value of 3Ͻ0 indicates a reduction in the cessfully withdraw from felbamate treatment (ie, felba- seizure rate. The correlations between the counts were mod- mate was discontinued or reduced, but seizure frequency eled as rij=a,iЈj(exchangeable correlations). Statistical sig- increased dramatically and the benefits of continuing fel- nificance was ␣Յ.05. Because there was a relatively small bamate outweighed the potential risks); (3) any progres- number of subjects, we examined each variable one at a time. sive neurologic problem with a propensity to increase Interactions were assessed first and when found not to be seizure frequency over time; or (4) participated in an significant (P>.05), each interaction was dropped and the investigational drug trial. model reassessed.

Table 1. Demographic Data*

Felbamate Felbamate All Patients No Gabapentin Gabapentin Responders Nonresponders Parameter (N = 45) (n = 18) (n = 27) (n = 14) (n = 31) Age, y 35 ± 11 34 ± 12 36 ± 10 35 ± 13 35 ± 10 Felbamate dose, mg/d 3028 ± 783 2578 ± 826 3181 ± 839 2807 ± 845 3129 ± 746 Felbamate use duration, mo 8 ± 3 7 ± 3 9 ± 3 9 ± 3 8 ± 3 Felbamate taper duration, wk 2.0 ± 0.6 2.0 ± 0.8 2.0 ± 0.5 2.0 ± 0.9 2.0 ± 0.4

*Data are given as mean ± SD.

Table 2. Doses and Concentrations of Concomitant Antiepileptic Drugs*

Baseline First Postfelbamate Month

PHT CBZ VPA PHT CBZ VPA Mean dose, mg/d 398 ± 132 (16) 1192 ± 385 (26) 1897 ± 791 (17) 310 ± 103 (16) 869 ± 315 (26) 1488 ± 856 (17) Mean concentration, µg/mL 14.8 ± 7.2 (13) 10.0 ± 2.3 (22) 71.3 ± 27.4 (16) 11.1 ± 5.9 (9) 9.3 ± 2.3 (8) 60.7 ± 28.2 (7)

*Data are given as mean ± SD (number of patients). Postfelbamate month indicates month after felbamate withdrawal; PHT, phenytoin sodium; CBZ, carbamazepine; and VPA, valproate sodium.

ARCH NEUROL / VOL 55, MAY 1998 642

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 In August 1994, the Food and Drug Administra- eventually received gabapentin had significantly more sei- tion issued 2 letters to physicians recommending ex- zures during the prefelbamate period (P=.03). treme caution when using felbamate, because of a pos- There was no difference in seizure frequency in the sible association between this AED and aplastic anemia prefelbamate and postfelbamate periods between fel- or hepatic failure. The actual risk of a potentially fatal bamate responders and nonresponders (P=.82). complication was unknown at that time, so we advised most patients in our center to discontinue felbamate COMMENT therapy. Soon after initiating a uniform withdrawal regimen in our patients, we began to see dramatic and pro- Most previous studies of AED drug discontinuation longed increases in seizure frequency. In addition, at least have examined short-term seizure frequency changes 1 report of status epilepticus during the tapering of fel- during inpatient presurgical evaluation. Several bamate appeared in the literature.4 We performed a ret- investigators6-8 found seizure rates were increased as rospective chart review to better characterize the nature much as 163% when AEDs were withdrawn over peri- and magnitude of seizure increases in our patients after ods of 10 days or less. A few studies have evaluated felbamate withdrawal. seizure frequency beyond the immediate withdrawal phase. Duncan et al9 found that the frequency of sei- RESULTS zures remained 66% above baseline 4 weeks after discontinuation of carbamazepine (rate of decrease, 200 A total of 95 patient charts were reviewed, with data from mg every 2 days), but withdrawal of phenytoin sodium 45 charts included in the analysis. Eight patients were and valproate sodium did not result in increased sei- excluded due to participation in investigational drug tri- zures. In a long-term study of barbiturate and benzodi- als, 1 because of felbamate-related hepatotoxic effects, azepine withdrawal, Theodore et al10 reported “tran- 25 due to a postfelbamate drug regimen that differed en- sient increases” for some patients in weekly seizure tirely from the prefelbamate regimen (1 patient experi- frequency during drug withdrawal, and only 1 of 78 enced status epilepticus during felbamate withdrawal), patients had more than a 50% increase in seizures at 3 because they received felbamate for less than 1 month, long-term follow-up (mean follow-up of 17 months for 5 due to an inability to discontinue felbamate, and 8 be- outpatients, 24 months for inpatients). In addition, cause of incomplete seizure calendars. Demographic data Doyle and colleagues,11 using standard AEDs, observed for patients included in the study are presented in a 45% decrease in seizure frequency when the previous Table 1. Doses and serum concentrations of the con- AED regimen was reestablished in patients who had comitant AED were slightly lower after felbamate with- medication withdrawn as part of a presurgical evalua- drawal compared with baseline doses and concentration of their seizures. The increase in seizure frequency tions (Table 2). we noted in the first month after felbamate withdrawal Seizure frequency increased significantly (P=.02) is similar to that seen with rapid withdrawal of other with felbamate withdrawal (Table 3). This effect peaked AEDs. In contrast, our patients had greater increases in at 106% above the mean baseline seizure rate during the seizure frequency beyond the immediate withdrawal month of felbamate tapering. The effect persisted for 3 phase than that reported in other studies. months after felbamate withdrawal with increased sei- Why was there a dramatic increase in seizure fre- zure rates ranging from 24% to 56% above the mean base- quency following felbamate withdrawal? The time line seizure frequency. The increase was independent of course of increased seizure frequency in our patients is age, sex, or maximum felbamate dose. beyond that expected for true withdrawal seizures Patients who received gabapentin had a smaller sei- produced when AED plasma concentrations are de- zure increase than those who did not receive gabapen- clining. The reasons for increased seizure frequency tin, when prefelbamate seizure frequency was com- with felbamate withdrawal are unclear and future pared with postfelbamate seizure frequency (Table 3). This investigation into the mechanism behind this phenom- difference did not reach statistical significance (P=.56). enon is indicated. Specific parameter estimates for this comparison are Patients whose postfelbamate AED regimen dif- shown in Table 4. These estimates allow solution of the fered substantially from that received before felbamate log-linear statistical equation and demonstrate a strong were carefully eliminated from our study. However, the trend toward gabapentin reducing the seizure increase mean doses of concomitant AED were lower during the seen with felbamate withdrawal. In addition, patients who postfelbamate period compared with the prefelbamate

Second Postfelbamate Month Third Postfelbamate Month Fourth Postfelbamate Month

PHT CBZ VPA PHT CBZ VPA PHT CBZ VPA 321 ± 101 (16) 912 ± 327 (26) 1503 ± 826 (17) 331 ± 98 (16) 888 ± 327 (26) 1476 ± 826 (17) 337 ± 98 (16) 885 ± 331 (26) 1450 ± 773 (17) 10.9 ± 3.8 (7) 9.8 ± 4.4 (3) 81.8 ± 27.8 (5) 10.2 ± 4.7 (8) 7.7 ± 1.4 (3) 55.3 ± 10.9 (4) 9.8 ± 3.3 (6) 7.1 ± 1.1 (5) 69.5 ± 20.8 (7)

ARCH NEUROL / VOL 55, MAY 1998 643

First Second Third Fourth Postfelbamate Postfelbamate Postfelbamate Postfelbamate Patient Group Baseline Month Month Month Month All patients 8.8 ± 13.8 18.2 ± 46.3 13.7 ± 27.6 10.9 ± 21.4 12.7 ± 25.2 Patients not receiving gabapentin 4.8 ± 5.7 9.3 ± 24.7 9.2 ± 24.7 9.4 ± 24.7 9.1 ± 24.7 Patients receiving gabapentin 11.5 ± 16.9 24.0 ± 56.2 16.7 ± 29.4 12.9 ± 20.2 15.1 ± 25.7

*Data are given as mean ± SD seizures per month. Postfelbamate month indicates month after felbamate withdrawal.

undergoing surgical evaluation or were discontinued from Table 4. Parameter Estimates ineffective medications. for Gabapentin vs No Gabapentin Gabapentin appeared to reduce the magnitude of the felbamate withdrawal effect. This effect did not reach sta- Equation Log Mean 95% Confidence Parameter Seizure Rate Interval P tistical significance, although the negative log of mean seizure rates (␤3) and the confidence interval range (Table ␤0 1.54 1.01 to 2.07 NA* 4) indicate a strong trend. Our small sample size and the ␤ 0.69 −0.36 to 1.73 .20 1 large variations among patients in seizure frequency dif- ␤2 0.91 0.14 to 1.68 .02

␤3 −0.32 −1.42 to 0.78 .57 ferences may have contributed to the lack of statistical significance. Patients who received gabapentin had a sig- *Not applicable. nificantly higher prefelbamate seizure frequency. We believe that this bias, introduced into the study by lack of period. We believe it is unlikely that the entire effect of randomization, was due to clinicians choosing to use gaba- increased seizures is due to lower doses of concomitant pentin, a new drug, in patients who appeared most re- AED, because the differences in doses were minor and fractory. A larger sample size and randomization would not statistically significant, whereas the magnitude of the allow a more thorough evaluation of this observation. seizure frequency increase was greater than changes re- By selecting patients who had more frequent sei- ported with most outpatient studies of complete AED zures to receive gabapentin, it could be argued that it was withdrawal. Furthermore, serum concentrations of the easier for patients not receiving gabapentin to experi- concomitant AED were lower during postfelbamate ence a doubling of their seizure frequency. However, dur- months than during the baseline. Serum concentrations ing the first month of felbamate withdrawal, while gaba- were measured in patients when there was a clinical in- pentin doses were being increased, the patients receiving dication (ie, increased seizure frequency or increased con- gabapentin also experienced a doubling of their seizure centration-related adverse effects). All concentrations were frequency. Subsequently, patients receiving gabapentin obtained due to increased seizure frequency or a recent had a decline in their seizure rates while those not re- dosage change, possibly selecting patients who were most ceiving gabapentin maintained a doubling of their base- likely to have decreased concentrations. Measured phen- line seizure rate. This effect may have led clinicians to ytoin and valproate concentrations remained relatively underestimate the efficacy of gabapentin. Gabapentin may stable throughout the 4 postfelbamate months and car- have had a positive therapeutic effect on seizure fre- bamazepine concentrations tended to decline, while the quency, but, because the clinician was unaware of the seizure frequency tended to decrease during the same pe- underlying withdrawal phenomenon, gabapentin may riod. This would indicate that serum concentrations were have been thought to be responsible for increasing not closely related to seizure frequency, because the larg- seizures. Without a prolonged trial of gabapentin, the est increases in seizures occurred during months with the patient may have been denied the use of a potentially highest serum concentrations. We believe it is unlikely effective drug. that increased seizure frequency was due to drug-drug These results also need to be considered in the de- pharmacokinetic interactions, because the AEDs used be- sign of AED trials. If, as observed with felbamate, an in- fore and after felbamate use were the same, except for crease in seizure frequency can continue for 3 or more the possible addition of gabapentin. Gabapentin has not months after AED discontinuation, a longer time may be been reported to interact with other AEDs. Neverthe- needed to assess seizure changes in studies using a cross- less, we cannot exclude the possibility that changes in over design. Second, in clinical trials comparing 2 or more postfelbamate AED doses and serum concentrations had AEDs after titration to monotherapy, the AED being with- some effect on the seizure frequency changes. drawn may have an important influence on subsequent A unique feature of our study is the minimization seizure frequency and the probability of the patient meet- of a bias present in previous reports, allowing a com- ing exit criteria while receiving monotherapy. parison of felbamate responders with nonresponders. We Limitations of this study include the retrospective discontinued felbamate regardless of the patient’s re- trial design, the fact that patients were not routinely re- sponse to the drug. Other studies tended to enroll pa- ceiving the same baseline dosage regimens, and that patients who did not respond to the medication being tients were not randomized in respect to receiving gaba- withdrawn, were refractory to AED therapy, and were pentin. Nevertheless, in many patients the effect on seizure

ARCH NEUROL / VOL 55, MAY 1998 644

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 frequency of discontinuing felbamate may be more in tense 3. Bromfield EB, Dambrosia J, Devinsky O, Nice FJ, Theodore WH. Phenytoin with- and last longer than generally appreciated. Clinicians and drawal and seizure frequency. Neurology. 1989;39:905-909. 4. DeGiorgio CM, Lopez JE, Lekht AN, Rabinowicz AL. Status epilepticus induced investigators should be aware that discontinuation of an by Felbatol withdrawal. Neurology. 1995;45:1021-1022. AED can have an important influence on the perceived 5. Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data. New York, NY: Oxford efficacy of any new treatment, whether a standard or Science Publications; 1995:146-168. new AED. 6. Malow BA, Blaxton TA, Stertz B, Theodore WH. Carbamazepine withdrawal: effects of taper rate on seizure frequency. Neurology. 1993;43:2280-2284. 7. Marks DA, Katz A, Scheyer R, Spencer SS. Clinical and electrographic effects Accepted for publication October 14, 1997. of acute anticonvulsant withdrawal in epileptic patients. Neurology. 1991;41: Reprints: Michael D. Privitera, MD, Department of Neu- 508-512. rology, University of Cincinnati Medical Center, ML 0525, 8. Spencer SS, Spencer DD, Williamson PD, Mattson RH. Ictal effects of anticon- 231 Bethesda Ave, Cincinnati, OH 45267-0525 (e-mail: vulsant medication withdrawal in epileptic patients. Epilepsia. 1981;21: 297-307. [email protected]). 9. Duncan JS, Shorvon SD, Trimble MR. Discontinuation of phenytoin, carbamazepine, and valproate in patients with active epilepsy. Epilepsia. 1990;31: REFERENCES 324-333. 10. Theodore WH, Porter RJ. Removal of sedative-hypnotic antiepileptic drugs from the regimens of patients with intractable epilepsy. Ann Neurol. 1983;13: 1. Theodore WH, Porter RJ, Raubertas RF. Seizures during barbiturate with- 320-324. drawal: relation to blood level. Ann Neurol. 1987;22:644-647. 11. Doyle WK, Devinsky O, Luciano D, Perrine K, Dogali M. Decreased seizure fre- 2. Schmidt D. Reduction of two-drug therapy in intractable epilepsy. Epilepsia. 1983; quency after withdrawal and reinstitution of antiepileptic drug therapy. Seizure. 24:368-376. 1994;3:61-65.

Announcement

Free Patient Record Forms Available Patient record forms are available free of charge to ARCHIVES readers by calling or writing FORMEDIC, 12D Worlds Fair Dr, Somerset, NJ 08873-9863, telephone (908) 469-7031.

ARCH NEUROL / VOL 55, MAY 1998 645