Epilepsy and Behavior 18 (2010) 74–80

Contents lists available at ScienceDirect

Epilepsy and Behavior

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / ye b e h

Cognitive outcome of antiepileptic treatment with levetiracetam versus

carbamazepine monotherapy: A non-interventional surveillance trial
Christoph Helmstaedter ⁎, Juri-Alexander Witt
University Clinic of Epileptology, University of Bonn, Bonn, Germany

a r t i c l e i n f o a b s t r a c t

Article history: This open-label, non-interventional, controlled surveillance study evaluated the cognitive outcome of
Received 19 February 2010 patients administered levetiracetam (LEV) or carbamazepine (CBZ) monotherapy as primary treatment or as
Received in revised form 22 February 2010 substitution for previous treatment. Executive functions, verbal memory, and subjective ratings were
Accepted 24 February 2010
assessed before and 6 months after initiation of LEV or CBZ monotherapy. Analyses included 498 patients:
Available online 13 May 2010
370 received LEV (63% pretreated), and 128 CBZ (34% pretreated). Mostly because of the substitution
condition, the seizure freedom rate was slightly higher with LEV as opposed to CBZ (78% vs 69%). Almost all
Keywords:
Levetiracetam
cognitive measures improved with LEV but not with CBZ, and repeated-measures MANOVA did not indicate
Carbamazepine seizure control or pretreatment as decisive with respect to cognitive change. With LEV, executive functions
Monotherapy improved in 15% and deteriorated in 5% of patients; the opposite pattern was seen under CBZ (improvement
Cognition with LEV OR = 2.3, deterioration with CBZ OR = 3.4). The findings suggest a mild but definitely superior
Neuropsychology cognitive outcome with LEV as opposed to CBZ monotherapy.
Quality of life © 2010 Elsevier Inc. All rights reserved.
Cognitive outcome
Memory
Executive functions
Epilepsy

1. Introduction
Levetiracetam (LEV) has been demonstrated to be efficacious in
the treatment of symptomatic and idiopathic epilepsy [1–5]. With LEV
monotherapy, 6-month seizure freedom was achieved by 73.0% of
patients with newly diagnosed epilepsy as compared with 72.8% on
carbamazepine (CBZ) monotherapy [6].
According to recent reviews, newer antiepileptic drugs (AEDs)
appear to be more favorable than the older drugs with respect to
cognitive side effects [7–9]. For LEV no negative cognitive side effects
have been described so far, and there are even studies suggesting that
LEV may have a stimulating effect on cognition. Two of five add-on
studies [10,11] indicated improved cognitive function, whereas the
other studies showed neutral effects [12–14]. A recent randomized,
double-blind, crossover study on LEV versus carbamazepine (CBZ) in
healthy volunteers revealed better results on half of the applied
neuropsychological measures under LEV as opposed to CBZ, whereas
none of the measures turned out to be superior under CBZ [15].
Another double-blind, placebo-controlled study with volunteers on
LEV versus CBZ versus oxcarbazepine (OXC) versus placebo reported a
neutral effect of LEV on reaction times, improved performance on the
⁎ Corresponding author. University Clinic of Epileptology Bonn, Sigmund Freud
Strasse 25, 53105 Bonn, Germany. Fax: +49 228 287 14486.
E-mail address: C.Helmstaedter@uni-bonn.de (C. Helmstaedter).
Stroop test for all three AEDs investigated with a pronounced effect of
LEV in particular, and psychomotor slowing under CBZ and OXC [16].
A recent study [10] that compared LEV with placebo indicated that
LEV has a positive effect on quality of life (QOL), which is in line with
previous reports of a positive effect of LEV on subjective outcome
measures [17,18].
On the basis of these findings, the first prospective, non-
interventional surveillance study in 2008 evaluated the cognitive
effects and efficacy of treatment with LEV in 401 outpatients seen by
practitioners [19]. LEV was given as initial treatment, as add-on
treatment, or as a substitute for the patient's antiepileptic medica-
tion. Repeated testing indicated cognitive improvement within the
6-month follow-up interval in 23–29% of the patients, whereas there
was deterioration in 5–6%. A positive effect of LEV apart from seizure
control on cognition could not be proven because of the lack of a
control condition, that is, an adequate comparator.
This second surveillance trial on LEV is a logical consequence of the
first study as it compares LEV with CBZ in the monotherapy condition.
The primary outcome parameter of this trial was cognition evaluated
at baseline and after a 6-month interval. In concordance with the
literature, the core hypothesis was that LEV should result in a superior
cognitive outcome as compared with CBZ. In particular, we expected
to confirm a positive effect of LEV on cognition apart from eventual
effects of seizure control, which were suggested by the findings of our
first trial. Seizure control was the secondary outcome parameter.

1525-5050/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.yebeh.2010.02.011
 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80
According to the literature, LEV should not be expected to be superior
to CBZ. Final data on efficacy, dropout rates and discontinuation, and
side effects will be presented in detail in a second report that will
cover the clinical epileptological data of all patients over the entire 1-
year observational period.
2. Methods
2.1. Study design
In a multicenter setting, physicians were asked to prescribe either
CBZ or LEV monotherapy to patients with newly diagnosed epilepsy or
patients being considered for a drug change. The choice of drug was
left to the doctors. Inclusion criteria were a diagnosis of epilepsy, age N
16, and cognitive assessment at baseline and at the 6-month follow-
up. The study protocol for the trial comprised evaluations at baseline,
a control visit for dose adjustment 1–4 months after inclusion, and
follow-up evaluations 6 and 12 months after inclusion. To study the
cognitive effects of the respective treatment conditions, about half of
the recruited patients were scheduled for neuropsychological assess-
ments before treatment and again after a 6-month interval. There was
no selection criterion for performing cognitive testing. The patients
included were evaluated with respect to seizures, medication,
cognition, and self-rated quality of life. Doctors were asked to provide
information about etiology, neuroimaging (MRI-revealed lesions),
and routine EEG (pathological EEG yes/no).
2.2. Study groups
A total of 988 patients were recruited for the trial on the efficacy of
LEV versus CBZ monotherapy. Initially, 588 of these 988 patients had
been recruited for the evaluation of cognitive outcomes after
6 months. However, 15% of the patients (65 in the LEV arm, 25 in
the CBZ arm) were lost to analysis because of missing test data at the
6-month follow-up. The final study comprised 498 patients (Table 1).
Two hundred seventy-six of these (55%) had already been taking
antiepileptic medication (pretreated), and it is important to note that
there were more pretreated patients in the LEV group (63%) than in
the CBZ group (34%) (χ² = 30.8, P b 0.001). As could be expected,
pretreated patients had an earlier age at onset and a longer duration of
epilepsy. Additionally, there were significantly more females in the
pretreated group. There were no significant group differences with
respect to age, education, type of epilepsy, or etiology (Table 2).
2.3. Outcome measures
2.3.1. Executive functions
The EpiTrack is a screening tool dedicated to the tracking of
adverse cognitive effects of antiepileptic medication. It assesses
attention and executive functions with six subtests, which result in
Table 1
Patients recruited for the cognitive part of the study.
LEV CBZ
Patients included in surveillance study 723 265
Pretreated 393 (59%) 82 (33%)
Non-pretreated 278 (41%) 163 (67%)
No information on pretreatment 72
Patients enrolled for cognitive 435 153
evaluation
Patients lost to follow-up cognitive 65 (15%) 25 (16%)
assessment
Patients included in cognitive 370 128
outcome evaluation
Pretreated 232 (63%) 44 (34%)
Non-pretreated 138 (37%) 84 (66%)
75
one test score [20]. Application and evaluation of this test are simple
and, thus, guarantee objectivity. Age-corrected norms and reliability
of change indices (RCIs) for retesting after 3–5 months are provided.
Patients can achieve a maximum score of 45 points. The interval for
mild impairment is 26–28 points. The cutoff for significant impair-
ment is ≤ 25 points. Changes as defined by practice-corrected
reliability of change intervals are significant at the 10% level with
gains N4 points, and losses N3 points.
2.4. Verbal learning and memory
As the EpiTrack assesses working memory and not episodic
memory, a shortened version of a German verbal learning and
memory test (VLMT) was used as well [21]. Memory testing consisted
of learning and immediate recall of a 15-item word list before the
EpiTrack. Delayed free recall of the learned items was requested after
the EpiTrack. The short version of the memory test was normalized
with 124 healthy subjects. After age correction, converted scores ≥ 5
were rated as normal, scores of 3–4 as mild impairment, and scores of
1–2 as significant impairment. Retest–reliability is not yet available. A
change was considered to be significant when patients moved from
one into another performance category.
2.4.1. Subjective measures (doctor rating)
1. At the first visit the attending physician had to rate cognitive status
(four levels: 1 = normal, 2 = slight impairment, 3 = moderate
impairment, 4 = strong impairment). A score N1 was considered
impairment
2. Six months after initiation of LEV or CBZ monotherapy the
attending doctor had to decide whether cognitive status had
changed since the first visit (five levels of cognitive change:
0 = no change, 1 = medium improvement, 2 = strong improve-
ment, –1 = medium deterioration, –2 strong deterioration).
Scores different from 0 were indicative of change.
2.4.2. Subjective measures (self-rating)
Prior to and 6 months after the initiation of LEV or CBZ
monotherapy a set of subjective measures were applied to assess
patients´ subjectively perceived cognitive status, QOL, and health
status in general.
1. Subjective overall health status (“health rating”) was rated on a
scale of 0–100 (absolute best to worst imaginable health status).
Scores b50 were considered as reflecting poorer health status.
2. QOL as assessed with the WHO-5 question questionnaire [22].
Scores N15 (range: 5–30) were indicative of impairment.
3. Subjective ratings focusing on:
a. “Cognitive capabilities” (three questions, impairment indicated
by a score N2)
b. “Daily life activities” (two questions, impairment indicated by a
score b5)
c. “AED tolerance” (two questions, impairment indicated by a
score N3)
Total
988 For these scales a significant change (improvement, deterioration)
475
between baseline and the 6-month follow-up was concluded if the
441
72 difference between the test times exceeded 1 SD in either direction. As
588 this study did not include an explicit mood scale, the WHO-5 QOL
measure was taken as a surrogate measure for depression if control
90 (15%) for depression was required [23].
498
2.4.3. Efficacy
276 Seizure outcome was calculated absolutely, that is, complete
222
seizure freedom since the control visit.
76 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80

Table 2
Characteristics of patients at baseline.

Total (N = 498) LEV (N = 370) CBZ (N = 128) Significance

Pretreated Non-pretreated Pretreated Non-pretreated

(N = 232) (N = 138) (N = 44) (N = 84)

Sex
a
Male 238 (48%) 97 (42%) 71 (51%) 20 (46%) 50 (60%)
Female 260 (52%) 135 (58%) 67 (49%) 24 (54%) 34 (40%)
Age, mean (SD) 46.2 (18.0) 45.2 (17.2) 47.3 (20.3) 42.6 (13.4) 48.8 (18.3) n.s.
Education N 10 years 127 (25.5%) 56 (24.1%) 43 (31.2%) 14 (31.8%) 14 (16.7%) n.s.
c
Age at onset of epilepsy, 40.8 (20.3) 37.7 (20.0) 46.1 (20.4) 31.6 (17.4) 45.8 (19.2)
mean (SD)
c
Duration of epilepsy years, 4.9 (9.3) 6.9 (10.6) 1.1 (3.6) 10.4 (12.0) 2.6 (7.4)
mean (SD)
Type of epilepsy:
Idiopathic. 156 (31%) 75 (32%) 40 (29%) 20 (46%) 21 (25%) n.s.
Cryptogenic 123 (25%) 59 (26%) 34 (25%) 8 (18%) 22 (26%)
Symptomatic 219 (44%) 98 (42%) 64 (46%) 16 (36%) 41 (49%)
Etiology
Vascular 95 (43%) 39 (40%) 35 (55%) 4 (25%) 17 (42%) n.s.
Trauma 45 (21%) 15 (15%) 13 (20%) 6 (38%) 11 (27%)
Tumor 26 (12%) 13 (13%) 7 (11%) 1 (6%) 5 (12%)
Infection 7 (3%) 5 (5%) 1 (1%) 1 (6%) 0 (0%)
Developmental 20 (9%) 14 (14%) 2 (3%) 1 (6%) 3 (7%)
Other 26 (12%) 12 (12%) 6 (9%) 3 (19%) 5 (12%)
Pathological EEG 287 (59%) [10 miss.] 139 (61%) 80 (60%) 24 (55%) 44 (52%) n.s.
Seizure days in last 6 months, 6.9 (18.8) 8.6 (22.9) 6.2 (17.5) 4.8 (5.7) 4.6 (11.4) n.s.
mean (SD)
AED treatment
Monotherapy 225 (82%) 184 (79%) — 41 (93%) — n.s.
Polytherapy 51 (18%) 48 (21%) — 3 (7%) —
Two AEDs 40 37 — 3 —
Three AEDs 11 11 — — —
Pre-treatment
Carbamazepine 91 (33%) 91 (33%) — — — L N Cc
Valproic acid 106 (38%) 83 (36%) — 23 (52%) — C N La
Lamotrigine 50 (18%) 42 (18%) — 8 (18%) — n.s.
Oxcarbazepine 19 (7%) 19 (8%) — — — L N Ca
Phenytoin 20 (7%) 16 (7%) — 4 (9%) — n.s.
Topiramate 16 (6%) 13 (6%) — 3 (7%) — n.s.
Gabapentin 10 (3.6) 10 (4%) — — — n.s.
Phenobarbital 11 (4%) 9 (4%) — 2 (5%) — n.s.
Other 12 (5%) 5 (2%) — 7 (15%) — C N Lb

C = Carbamazepine group; L = Levetiracetam group.

a
P b 0.05.
b
P b 0.01.
c
P b 0.001.

2.5. Statistical analysis
Frequency statistics for categorical data and odds ratios were
calculated with χ² tests. Inference statistics for baseline data were
calculated with ANOVA. Changes in dependent measures over time as
a function of “cohort” (LEV vs CBZ), “pretreatment” (yes vs no), and
“seizure outcome” (seizure free vs seizures) were calculated with
repeated-measures MANOVA.
3. Results
3.1. Baseline conditions (T1)
Table 3 summarizes the categorized objective and subjective
performance measures for pretreated and non-pretreated patients
administered CBZ or LEV. When both mild impairment and significant
impairment were taken into account, there was subjective impairment
in up to 66% of all patients and objective cognitive impairment in up to
62% (Table 3). There were no baseline differences in objective cognitive
measures between the four treatment groups. Among the subjective
measures, only doctors’ ratings differed in that pretreated patients
given LEV were more frequently rated as impaired than patients in the
other groups. (χ² = 17.7, P = 0.007). All subjective ratings were
significantly correlated with each other (r ranging from 0.32 to 0.57,
P b 0.001). The highest correlation (r = 0.57, P b 0.001) was observed
between patients’ ratings of cognition and the QOL measure. Good
correlations were also observed between doctors’ ratings of patients’
cognition and other cognition-related subjective and objective out-
come measures (memory: r = 0.32, EpiTrack: r = 0.44, patients’ rating
of cognition: r = 0.24). However, doctors were most likely not blind to
test results or complaints about cognitive problems. Correlations
between patients’ subjective ratings and the results of objective testing
were less pronounced, and reached statistical significance mainly
because of the large sample size (subjective ratings of cognition: r =
0.14 for memory and 0.20 for EpiTrack; QOL: r = 0.09 for EpiTrack).
Type of epilepsy affected objective (EpiTrack: F = 13.2, P b 0.001;
verbal learning: F = 19, P b 0.001) and subjective (cognitive status:
F = 12.5, P b 0.001; everyday life activities: F = 6.5, P = 0.002) baseline
scores. Patients with symptomatic epilepsy performed worse and
were rated more poorly than those with idiopathic or cryptogenic
epilepsy.
3.2. Outcome at the 6-month follow-up (T2)
3.2.1. Titration of LEV and CBZ
The control visit took place after 1.8 ± 1.1 months, and the 6-
month follow-up (T2) after 5.7 ± 0.9 months. In non-pretreated
patients, the intended monotherapy started on day 1. In pretreated
 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80 77

Table 3
Subjective and objective performance measures at baseline.

Total (N = 498) LEV (N = 370) CBZ (N = 128) Significance

Pretreated Non-pretreated Pretreated Non-pretreated

(N = 232) (N = 138) (N = 44) (N = 84)

Health rating (1–100)

Impaired (N = 459) 142 (31%) 75 (34%) 31 (25%) 14 (33%) 22 (29%) n.s.
Quality of life (patient)
Impaired (N = 476) 283 (60%) 137 (61%) 75 (58%) 27 (61%) 44 (55%) n.s.
Ratings of cognition, activities, and AED tolerance (patient)
Impaired cognition (N = 457) 232 (51%) 113 (52%) 62 (50%) 24 (56%) 33 (46%) n.s.
a
Impaired activities (N = 456) 300 (66%) 136 (63%) 76 (62%) 35 (81%) 53 (71%)
Impaired by AED (N = 238) pretreated 95 (40%) 80 (40%) — 15 (39%) — n.s.
Rating of cognitive performance (doctor)
b
No impairment 282 (57%) 111 (48%) 96 (70%) 24 (55%) 51 (61%)
Mild impairment 146 (29%) 83 (36%) 29 (21%) 12 (27%) 22 (26%)
Impairment 69 (14%) 37 (16%) 13 (9%) 8 (18%) 11 (13%)
EpiTrack
No impairment 230 (46%) 103 (44%) 65 (47%) 22 (50%) 40 (48%) n.s.
Mild impairment 77 (16%) 29 (13%) 27 (20%) 4 (9%) 17 (20%)
Impairment 191 (38%) 100 (43%) 46 (33%) 18 (41%) 27 (32%)
Verbal learning
No impairment 188 (38%) 92 (40%) 47 (35%) 19 (43%) 30 (36%) n.s.
Mild impairment 252 (51%) 115 (50%) 66 (49%) 24 (55%) 47 (57%)
Impairment (7 missing) 51 (11%) 22 (10%) 22 (16%) 1 (2%) 6 (7%)
Verbal memory
No impairment 321 (68%) 150 (67%) 90 (70%) 28 (65%) 53 (68%) n.s.
Mild impairment 26 (6%) 10 (5%) 11 (9%) 3 (7%) 2 (2%)
Impairment (25 missing) 126 (26%) 62 (28%) 29 (22%) 12 (28%) 23 (30%)
a
P b 0.05.
b
P b 0.01.

patients, conversion to LEV monotherapy lasted 3.5 ± 4 weeks, and
conversion to CBZ monotherapy, 3.0 ± 3.1 weeks.
The mean CBZ doses for the pretreated and non-pretreated patients
were 895 ± 300 and 717 ± 300 mg/day at the control visit and 955 ±
332 and 789 ± 357 mg/day at the 6-month follow-up, respectively. The
respective doses for LEV were 1486 ± 570 (pretreated) and 1261± 460
(non-pretreated) mg/day at the control visit and 1568 ± 588 (pre-
treated) and 1311 ± 500 (non-pretreated) mg/day at T2. Doses were
higher for pretreated patients given either drug (F = 6.6–18.3, P = 0.01–
0.001) Doses were significantly increased after the control visit (CBZ:
T = 4.2, LEV: T = 4.8, P b 0.001 each), particularly when seizures were
not yet controlled. LEV dose increased by 177 ± 383 mg/day for those
with seizures versus 29± 209 mg/day for those who were seizure free,
and CBZ dose increased by 151 ± 251 mg/day for those with seizures
versus 27 ± 115 mg/day for those who were seizure free (P = 0.02–
0.001).
3.2.2. Efficacy
At the 6-month follow-up, 78% of the patients treated with LEV
and 69% of the patients taking CBZ had been seizure free since the
control visit (χ² = 4.26, P = 0.04). Seizure freedom rates for pretreated
and non-pretreated patients did not differ in the LEV arm (77% for
pretreated vs 80% for non-pretreated). However, pretreatment made
a significant difference with respect to seizure outcome in the CBZ
arm (57% for pretreated vs 75% for non-pretreated; χ² = 9.8, P = 0.02).
When the seizure freedom rates for pretreated patients given LEV or
CBZ were compared, the odds ratio of becoming seizure free with LEV
was 2.5 (CI = 1.3–4.9, χ² = 7.5, P = 0.0007). The difference in non-
pretreated patients was not significant (χ² = 0.67, P = 0.255).
Efficacy of LEV or CBZ did not differ among the symptomatic,
idiopathic, and cryptogenic epilepsies. Seizure freedom rates under
CBZ were 68% for idiopathic epilepsy (LEV 78%), 60% for cryptogenic
epilepsy (LEV 76%), and 63% for symptomatic epilepsy (LEV 78%).
Seizure outcome also did not differ when the 41 CBZ-treated patients
with idiopathic epilepsy were subdivided into those with focal
(N = 23) and those with generalized (N = 18) epilepsy (χ² = 0.39,
P = 0.65). This finding is important as idiopathic generalized epilepsy
may worsen under CBZ and some have suggested that CBZ should be
contraindicated in this type of epilepsy [24].
3.2.3. Cognitive outcome
Changes in dependent measures over time were evaluated with
repeated-measures MANOVA as a function of “cohort” (LEV vs CBZ),
“pretreatment” (yes vs no), and “seizure outcome” (seizure freedom
vs seizures). The values for the different dependent variables showed
that in most cases the “retest × cohort” interaction was significant.
There were no significant interaction effects including “pretreat-
ment,” and the “retest × seizure” interaction was significant only with
respect to subjective variables.
In detail, significant improvements in attention and executive
functions as assessed with the EpiTrack were seen only in the group
treated with LEV; the CBZ-treated group showed no significant change
in performance (“retest × cohort” interaction: F = 12.4, P b 0.001). Both
pretreated (T = –7.8, P b 0.001) and non-pretreated (T = –4.1, P b 0.001)
patients improved under LEV. The significant “test × cohort” interaction
for verbal learning and memory (F = 3.8, P = 0.02) was due mainly to
improvements in verbal learning in the group treated with LEV (F = 6.8
P = 0.009). As depicted in Fig. 1, effect sizes (Cohen's d) of the observed
improvements in objective cognitive measures under LEV ranged from
0.16 (verbal learning) to 0.21 (attention and executive functions).
For the combined analysis of “subjective cognitive impairment”
and “subjectively impaired activities” (F = 10.8, P b 0.001), the inter-
action became significant because patients taking LEV rated their
cognitive status as being better than before (F = 3.3, P b 0.001). An
interaction effect was not observed for patients’ daily activity ratings.
The “test × cohort” interaction (F = 5.4, P = 0.004) in the combined
analysis of subjective “health rating” and QOL indicated greater
improvement with LEV. In addition, the “test × seizure outcome”
interaction (F = 4.0, P = 0.01) was significant; that is, seizure-free
patients reported greater improvement in health and QOL than
patients with persistent seizures.
Pretreated patients had improved AED tolerance ratings with LEV
(“test × cohort” interaction: F = 10.1, P = 0.002). Seizure freedom
improved AED tolerance ratings as well (significant “seizure freedom ×
78 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80

Fig. 1. Effect sizes and 95% confidence intervals for the changes in the dependent measures. Positive effect sizes indicate improvement; negative effect sizes, deterioration.

test” interaction: F = 8.8, P = 0.003). The effects were generally more
pronounced on subjective than on objective measures (Fig. 1).
When the data were evaluated on the individual level, the four
treatment groups differed in the frequencies of individual categorical
changes in many respects (Table 4). The EpiTrack data indicate
improvement in performance in about 14–16% and deterioration in
performance in 4–6% of patients receiving LEV, whereas the opposite
pattern of 2–8% (improvement) and 11–17% (deterioration) was
observed in patients on CBZ (OR for improvement with LEV = 2.3, 95%
CI = 1.1–4.7; OR for deterioration with CBZ = 3.4, 95% CI = 1.7–6.7).
Doctors’ ratings of changes in patients’ cognitive performance
were largely consistent with the EpiTrack results: ratings of cognitive
improvement were more frequent under LEV (pretreated: 51%, non-
pretreated: 33%) as opposed to CBZ (pretreated: 5%, non-pretreated:
11%) treatment, and ratings of cognitive deterioration were more
frequent with CBZ (pretreated: 19%, non-pretreated: 11%) as opposed
to LEV (pretreated: 1%, non-pretreated: 2%) treatment (OR for
improvement under LEV = 5.5, 95% CI = 3.1–10.1). In addition,
patients themselves rated their cognition as improved under LEV
(pretreated: 59%, non-pretreated: 62%) more frequently than under
CBZ (pretreated: 38%, non-pretreated: 39%) (pretreated: χ2 = 6.0,
P = 0.01; nonpretreated: χ2 = 9.3, P = 0.002).
All groups rated overall QOL as improved, but again improvement
was more frequently observed in those treated with LEV (OR = 2.3,
95% CI = 1.3–4.1).
Individual changes in learning or memory did not differ between
the treatment groups (Table 4).
4. Discussion
In line with reports in the literature of positive effects of LEV
treatment on cognition, the major aim of the present study was to
demonstrate the suggested positive cognitive effects of LEV under
controlled conditions by contrasting LEV monotherapy with CBZ
monotherapy. The study was the direct and logical consequence of a
previous uncontrolled surveillance trial on LEV that indicated that
cognition improved when LEV was the first treatment, add-on
treatment, or substitution treatment [19].
4.1. Baseline conditions
A large cohort of 588 outpatients considered for treatment with
LEV or CBZ monotherapy were recruited for the evaluation of
cognitive outcome. Because of the dropout rate of 15%, 498 patients

Table 4
Individual changes in subjective and objective categorical measures.

LEV (N = 370) CBZ (N = 128) Significance

Pretreated Non-pretreated Pretreated Non-pretreated

↑a ↓ ↑ ↓ ↑ ↓ ↑ ↓
c
EpiTrack (498 datasets) 13.8% 4.3% 15.9% 5.8% 2.3% 11.4% 8.3% 16.7%
Learning (488 datasets) 19.4% 12.8% 20.0% 10.4% 6.8% 15.9% 13.4% 14.6% n.s.
Memory (465 datasets) 19.0% 14.8% 14.0% 17.8% 11.6% 14.0% 16.9% 10.4% n.s.
d
Doctor rating (483 datasets) 51.1% 0.9% 33.3% 1.5% 4.7% 18.6% 10.8% 10.8%
Global Health (443 datasets) 35.8% 2.4% 34.7% 4.2% 24.4% 4.9% 23.6% 5.6% n.s.
b
QOL (449 datasets) 39.5% 3.8% 31.7% 4.9% 16.3% 7.0% 21.9% 1.4%
c
Cognition (434 datasets) 58.8% 15.1% 62.2% 19.3% 38.1% 33.3% 39.1% 26.1%
Activities (432 datasets) 42.4% 1.5% 39.3% 0.9% 42.9% 7.1% 36.7% 4.3% n.s.
AED tolerance (222 datasets) 43.0% 3.2% 27.8% 2.8% n.s.
a
↓, significant decline; ↑, significant improvement.
b
P b 0.05.
c
P b 0.01.
d
P b 0.001.
 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80
remained for the final analysis. The greater number of pretreated
patients receiving LEV (63% vs 34% CBZ) may place this group at a
disadvantage as these patients already had failed to respond to their
first medication and therefore could be assumed to be less likely to
become seizure free [25,26]. With the exception of the earlier age at
onset, longer duration of epilepsy, and larger number of women than
men in the pretreated group, the treatment cohorts were comparable
with respect to clinical or demographic aspects.
At baseline, subjective and objective impairments in cognition,
QOL, and daily activities were evident in up to 66% of the patients.
Pretreated patients were in part rated as more severely impaired than
non-pretreated patients, but this result did not hold true with
objective testing. The fact that cognitive impairments are common
in newly diagnosed patients is in line with data recently published by
Taylor and colleagues, who compared cognitive profiles of newly
diagnosed patients with epilepsy with those of healthy control
subjects [27]. Fifty-three percent of the patients were significantly
impaired on at least one cognitive measure. The finding that patients
with symptomatic epilepsy performed worse than patients with
idiopathic epilepsy is also consistent with the literature [28].
Although subjective measures addressed different issues and
although different scales and ratings were used, the subjective measures
were highly interrelated and thus did not appear to assess independent
behavioral features. The QOL measure, taken as a surrogate measure for
depression, was highly correlated with patients´ complaints about their
cognitive status [23,29]. Consequently, patients’ reports in this study
may be considered as reflecting nonspecific mood rather than
qualitatively different aspects of well-being or impairment.
4.2. Follow-up evaluation
In concordance with the protocol, the control visit took place about
2 months after initiation of LEV or CBZ monotherapy. Doses were
increased before as well as after the control visit, particularly when
patients were not seizure free. At the time of the final visit (T2), the
mean doses of LEV (1470 ± 571 mg/day) and CBZ (846 ± 313 mg/day)
were within the midrange of the recommended doses of 1000–
3000 mg/day for LEV and 400–1200 mg/day for CBZ.
The explicit aim of this surveillance trial was to show a superior
cognitive outcome with LEV monotherapy as opposed to CBZ
monotherapy. This was confirmed by almost all objective and
subjective outcome measures. Verbal memory in terms of the loss of
learned words over time was the only exception. However, the
treatment arms differed with respect to the proportion of pretreated
patients at baseline, and the different treatment conditions resulted in
different seizure outcomes. Accordingly, pretreatment and seizure
control had to be considered factors that potentially influenced
cognition in addition to the chosen antiepileptic drug. Thus, the
secondary outcome parameter “seizure control” is discussed first.
Seizure outcome tended to be favorable for the cohort treated with
LEV. Seventy-eight percent of LEV-treated patients, in contrast to 69%
of CBZ-treated patients, became seizure free. However, closer analysis
revealed that this result was due mainly to the poor outcome in the
pretreatment condition of the CBZ arm, in which only 57% of patients
became seizure free (LEV 77%). In non-pretreated patients prescribed
CBZ monotherapy, the seizure freedom rate was much better (75%)
and not different from that of LEV (80%). Although it is not surprising
that seizure control is worse in the substitution (conversion)
condition when a previous drug has already failed, the lack of this
difference in the LEV arm was unexpected. Although remaining
speculative, perhaps this result is due to a different mechanism of
action of LEV in comparison to the premedication [30–32]. In the
pretreatment condition, 95% of the patients in the LEV arm and 98% of
the patients in the CBZ arm had been pretreated with AEDs targeting
sodium channels. Another reason for the poor seizure outcome with
CBZ in the substitution condition may be the large proportion of
79
patients with idiopathic epilepsy (45%) in that group. Treatment of
idiopathic epilepsy with CBZ can aggravate primary generalized
epilepsy [24], but this was not confirmed by the present data. With
respect to non-pretreated patients, the results confirm the equivalent
efficacy of LEV and CBZ that had been demonstrated by Brodie and
colleagues in 2007 [6]. Remission rates (LEV: 80% vs CBZ: 75%) were
slightly better than those reported in that study for LEV (73%) and
controlled-release CBZ (72.8%). Summarizing the findings on efficacy,
superior efficacy of LEV compared with CBZ can be concluded only for
the conversion condition when these drugs are substituted for
previous monotherapy.
In the previous surveillance trial [19], a combined positive effect of
LEV and seizure control on cognition was assumed. Improvement
under LEV independent of seizure control was observed within the
first 3 months, and only seizure-free patients showed further
improvement after this time. In the present analysis, better outcomes
were observed under LEV as opposed to CBZ which appeared to be
independent of treatment condition (pretreated) and seizure out-
come (seizure freedom). It may thus be concluded from the present
data that LEV has a positive psychotropic effect on cognition apart
from cognitive release effects resulting from seizure control or
substitution of medication.
Improvement was noted on almost all measures of cognition with
LEV, but not with CBZ. Health, QOL, and activity ratings improved
under CBZ as well, although these improvements were less pro-
nounced when compared with those resulting from LEV. A positive
change in attention and executive function, rather than memory,
points to a nonspecific activating effect of LEV on cognition. This effect
parallels the positive activating effects of LEV on self-reported
behavior when LEV efficiently controls seizures [33]. Both subjective
and objective measures demonstrated the same tendencies, but
generally, the effect sizes for subjective measures were larger than
those for objective measures.
Individual performance changes as assessed with EpiTrack are of
particular interest because they indicated improvement with LEV and
deterioration with CBZ. The odds ratio for cognitive improvement
with LEV was 2.3, and the odds ratio for deterioration with CBZ, 3.4.
Categorical changes in doctors’ ratings of cognition showed the same
tendency, but whether the doctors knew the results from objective
testing cannot be excluded. Finally, patients more frequently rated
their cognition and QOL as improved with LEV than with CBZ.
5. Conclusion
This large non-interventional study indicates that compared with
CBZ, LEV has slightly but definitely superior effects on cognition in
both pretreated and non-pretreated patients. This is reflected in
objective and subjective measures of cognition as well. Thus, in
answer to the open question in the uncontrolled surveillance trial
from 2008, LEV indeed has a positive stimulating effect on cognition
apart from the effect of seizure control or drug change. As for seizure
freedom, the results confirm that LEV and CBZ are equally effective as
monotherapy in non-pretreated patients. In the conversion (pre-
treated) condition, better seizure control is observed with LEV.
The shortcomings of this investigation are common for this type of
study (it was a nonrandomized, open-label, non-interventional,
multicenter study). The large sample size, the naturalistic clinical
outpatient setting, and the sensitive objective and subjective outcome
measures may be seen as strengths of this study. When both cognition
and seizure outcome results are taken into consideration, the results
favor LEV monotherapy over CBZ monotherapy, particularly when
LEV substitutes a preexisting therapy.
Conflict of interest statement
C. Helmstaedter received honoraria and a license fee from UCB.
80 C. Helmstaedter, J.-A. Witt / Epilepsy and Behavior 18 (2010) 74–80
Acknowledgments
This comparative surveillance study was sponsored by UCB GmbH,
Germany.
Data entry and preselection of data sets for the per-protocol analysis
were done by Kiencke P, Research & Public Relations, Burscheid, Germany.
References
[1] Ben-Menachem E, Falter U. for the European Levetiracetam Study Group. Efficacy
and tolerability of levetiracetam 3000 mg/d in patients with refractory partial
seizures: a multicenter, double-blind, responder-selected study evaluating
monotherapy. Epilepsia 2000;41:1276–83.
[2] Shorvon SD, Lowenthal A, Janz D, Bielen E, Loiseau P. for the European
Levetiracetam Study Group. Multicenter double-blind, randomized, placebo-
controlled trial of levetiracetam as add-on therapy in patients with refractory
partial seizures. Epilepsia 2000;41:1179–86.
[3] Cereghino JJ, Biton V, Abou-Khalil B, Dreifuss F, Gauer LJ, Leppik I. Levetiracetam
for partial seizures: results of a double-blind, randomized clinical trial. Neurology
2000;55:236–42.
[4] Berkovic SF, Knowlton RC, Leroy RF, Schiemann J, Falter U. Placebo-controlled study
of levetiracetam in idiopathic generalized epilepsy. Neurology 2007;69:1751–60.
[5] Noachtar S, Andermann E, Meyvisch P, Andermann F, Gough WB, Schiemann-
Delgado J. Levetiracetam for the treatment of idiopathic generalized epilepsy with
myoclonic seizures. Neurology 2008;70:607–16.
[6] Brodie MJ, Perucca E, Ryvlin P, Ben-Menachem E, Meencke HJ. Comparison of
levetiracetam and controlled-release carbamazepine in newly diagnosed epilepsy.
Neurology 2007;68:402–8.
[7] Aldenkamp AP, De Krom M, Reijs R. Newer antiepileptic drugs and cognitive
issues. Epilepsia 2003;44(Suppl 4):21–9.
[8] Meador KJ. Cognitive and memory effects of the new antiepileptic drugs. Epilepsy
Res 2006;68:63–7.
[9] Ortinski P, Meador KJ. Cognitive side effects of antiepileptic drugs. Epilepsy Behav
2004;5(Suppl 1):S60–5.
[10] Zhou B, Zhang Q, Tian L, Xiao J, Stefan H, Zhou D. Effects of levetiracetam as an add-
on therapy on cognitive function and quality of life in patients with refractory
partial seizures. Epilepsy Behav 2008;12:305–10.
[11] Piazzini A, Chifari R, Canevini MP, Turner K, Fontana SP, Canger R. Levetiracetam:
an improvement of attention and of oral fluency in patients with partial epilepsy.
Epilepsy Res 2006;68:181–8.
[12] Ciesielski AS, Samson S, Steinhoff BJ. Neuropsychological and psychiatric impact of
add-on titration of pregabalin versus levetiracetam: a comparative short-term
study. Epilepsy Behav 2006;9:424–31.
[13] Gomer B, Wagner K, Frings L, et al. The influence of antiepileptic drugs on cognition: a
comparison of levetiracetam with topiramate. Epilepsy Behav 2007;10:486–94.
[14] Neyens LG, Alpherts WC, Aldenkamp AP. Cognitive effects of a new pyrrolidine
derivative (levetiracetam) in patients with epilepsy. Prog Neuropsychopharmacol
Biol Psychiatry 1995;19:411–9.
[15] Meador KJ, Gevins A, Loring DW, et al. Neuropsychological and neurophysio-
logic effects of carbamazepine and levetiracetam. Neurology 2007;69:2076–84.
[16] Mecarelli O, Vicenzini E, Pulitano P, et al. Clinical, cognitive, and neurophysiologic
correlates of short-term treatment with carbamazepine, oxcarbazepine, and
levetiracetam in healthy volunteers. Ann Pharmacother 2004;38:1816–22.
[17] Cramer JA, Van Hammee G. Maintenance of improvement in health-related
quality of life during long-term treatment with levetiracetam. Epilepsy Behav
2003;4:118–23.
[18] French J, Edrich P, Cramer JA. A systematic review of the safety profile of
levetiracetam: a new antiepileptic drug. Epilepsy Res 2001;47:77–90.
[19] Helmstaedter C, Witt JA. The effects of levetiracetam on cognition: a non-
interventional surveillance study. Epilepsy Behav 2008;13:642–9.
[20] Lutz MT, Helmstaedter C. EpiTrack: Tracking cognitive side effects of medication
on attention and executive functions in patients with epilepsy. Epilepsy Behav
2005;7:708–14.
[21] Helmstaedter C, Lendt M, Lux S. VMLT Verbaler Lern- und Merkfähigkeitstest.
Göttingen: Beltz Test GmbH; 2001.
[22] Folker H, Folker AP. WHO-5 as a simple method for measuring quality of life in
daily psychiatric clinics. Ugeskr Laeger 2008;170:830–4.
[23] Sawrie SM, Martin RC, Kuzniecky R, et al. Subjective versus objective memory
change after temporal lobe epilepsy surgery. Neurology 1999;53:1511–7.
[24] Genton P. When antiepileptic drugs aggravate epilepsy. Brain Dev 2000;22:75–80.
[25] Kwan P, Brodie MJ. Effectiveness of first antiepileptic drug. Epilepsia 2001;42:
1255–60.
[26] Kwan P, Brodie MJ. Epilepsy after the first drug fails: substitution or add-on?
Seizure 2000;9:464–8.
[27] Taylor J, Kolamunnage-Dona R, Marson AG, Smith PE, Aldenkamp AP, Baker GA.
Patients with epilepsy: cognitively compromised before the start of antiepileptic
drug treatment? Epilepsia 2010;51:48–56.
[28] Loring DW, Barr W, Hamberger M, Helmstadter C. Neuropsychology evaluation -
adults. In: Engel JJ, Pedley TA, editors. Epilepsy: a comprehensive textbook.
Philadelphia: Wolters Kluwer/Lippincott Williams & Wilkins; 2008. p. 1077–84.
[29] Hoppe C, Elger CE, Helmstaedter C. Long-term memory impairment in patients
with focal epilepsy. Epilepsia 2007;48(Suppl 9):26–9.
[30] Lukyanetz EA, Shkryl VM, Kostyuk PG. Selective blockade of N-type calcium
channels by levetiracetam. Epilepsia 2002;43:9–18.
[31] Madeja M, Margineanu DG, Gorji A, et al. Reduction of voltage-operated potassium
currents by levetiracetam: a novel antiepileptic mechanism of action? Neuro-
pharmacology 2003;45:661–71.
[32] Rigo JM, Hans G, Nguyen L, et al. The anti-epileptic drug levetiracetam reverses the
inhibition by negative allosteric modulators of neuronal GABA- and glycine-gated
currents. Br J Pharmacol 2002;136:659–72.
[33] Helmstaedter C, Fritz NE, Kockelmann E, Kosanetzky N, Elger CE. Positive and
negative psychotropic effects of levetiracetam. Epilepsy Behav 2008;13:
535–41.