Tolerability, safety, and side effects of levetiracetam versus phenytoin in intravenous and total prophylactic regimen among craniotomy patients: A prospective randomized study


  • Karen L. Fuller,

    1. Centre for Clinical Neurosciences and Neurological Research, St Vincent’s Hospital, Melbourne, Victoria, Australia
    Search for more papers by this author
  • Yi Yuen Wang,

    1. Centre for Clinical Neurosciences and Neurological Research, St Vincent’s Hospital, Melbourne, Victoria, Australia
    Search for more papers by this author
  • Mark J. Cook,

    1. Centre for Clinical Neurosciences and Neurological Research, St Vincent’s Hospital, Melbourne, Victoria, Australia
    2. Department of Medicine, St Vincent’s Hospital Melbourne, The University of Melbourne, Melbourne, Victoria, Australia
    Search for more papers by this author
  • Michael A. Murphy,

    1. Centre for Clinical Neurosciences and Neurological Research, St Vincent’s Hospital, Melbourne, Victoria, Australia
    Search for more papers by this author
  • Wendyl J. D’Souza

    1. Centre for Clinical Neurosciences and Neurological Research, St Vincent’s Hospital, Melbourne, Victoria, Australia
    2. Department of Medicine, St Vincent’s Hospital Melbourne, The University of Melbourne, Melbourne, Victoria, Australia
    Search for more papers by this author

Address correspondence to Karen Fuller, Centre for Clinical Neurosciences and Neurological Research, Level 5 Daly Wing, St Vincent’s Hospital, 35 Victoria Parade, Melbourne, Fitzroy 3065, Vic., Australia. E-mail:


Purpose:  Practical choice in parenteral antiepileptic drugs (AEDs) remains limited despite formulation of newer intravenous agents and requirements of special patient groups. This study aims to compare the tolerability, safety, and side effect profiles of levetiracetam (LEV) against the standard agent phenytoin (PHT) when given intravenously and in total regimen for seizure prophylaxis in a neurosurgical setting.

Methods:  This prospective, randomized, single-center study with appropriate blinding comprised evaluation pertaining to intravenous use 3 days following craniotomy and at discharge, and to total intravenous-plus-oral AED regimen at 90 days. Primary tolerability end points were discontinuation because of side effect and first side effect. Safety combined end point was major side effect or seizure. Seizure occurrence and side effect profiles were compared as secondary outcomes.

Key Findings:  Of 81 patients randomized, 74 (36 LEV, 38 PHT) received parenteral AEDs. No significant difference attributable to intravenous use was found between LEV and PHT in discontinuation because of side effect (LEV 1/36, PHT 2/38, p = 1.00) or number of patients with side effect (LEV 1/36, PHT 4/38, p = 0.36). No significant difference was found between LEV and PHT total intravenous-plus-oral regimen in discontinuation because of side effect (hazard ratio [HR] 0.78, 95% confidence interval [CI] 0.21–2.92, p = 0.72) or number of patients with side effect (HR 1.51, 95% CI 0.77–2.98, p = 0.22). More patients assigned PHT reached the undesirable clinical end point for safety of major side effect or seizure (HR 0.09, 95% CI 0.01–0.70, p = 0.002). Seizures occurred only in patients assigned PHT (n = 6, p = 0.01). Although not significant, trends were observed for major side effect in more patients assigned PHT (p = 0.08) and mild side effect in more assigned LEV (p = 0.09).

Significance:  Both LEV and PHT are well-tolerated perioperatively in parenteral preparation, and in total intravenous-plus-oral prophylactic regimen. Comparative safety and differing side effect profile of intravenous LEV supports use as an alternative to intravenous PHT.

Impetus to optimize individual therapy is served by a wide and expanding range of oral antiepileptic drugs (AEDs), but clinical studies are relatively limited for settings requiring intravenous AEDs. Parenteral formulations include phenytoin (PHT), fosphenytoin (fosPHT), levetiracetam (LEV), valproate, lacosamide, benzodiazepines (BDZs), and anesthetic agents. Practical choice of intravenous AEDs is limited by availability dependent on national guidelines, and hospital formulary inclusion, where intravenous PHT has been standard with intravenous fosPHT often discounted on cost (Browne, 1998). Comparative safety and efficacy data is necessary to direct best use of newer intravenous AEDs.

The ideal intravenous AED is rapidly highly effective, safe and well tolerated, does not interact significantly with other medications or require frequent serologic monitoring, is affordable (Mattson, 1996; DeToledo & Ramsay, 2000), and appropriate to continue or resume orally. Intravenous PHT is effective in most settings but has a high side effect frequency and profile of serious cardiac and local complications (Earnest et al., 1983; Mattson, 1996). Inherent factors including nonlinear pharmacokinetics and alkaline formulation (Mattson, 1996; Martinelli & Muhlebach, 2003) mandate frequent monitoring, compromise tolerability, and limit administration rate thereby increasing care cost (Browne, 1998; Martinelli & Muhlebach, 2003).

Intravenous LEV presents an alternative as a potent newer intravenous AED, which is not hepatically metabolized, with little potential for local or venous access complication and a comparatively wide therapeutic window not requiring slow titration or serologic monitoring at standard doses (Ramael et al., 2006). LEV is highly effective as therapy for both focal and primarily generalized epilepsy with novel mode of action binding to synaptic vesicle protein 2A, linear kinetics, few serious side effects and minimal drug-drug interaction, although utility in early status epilepticus (SE) is theoretically limited by comparatively slower brain entry (Crepeau & Treiman, 2010). Intravenous LEV is in off-label use in patient groups potentially more suited to its pharmacokinetics and favorable tolerability (Ramael et al., 2006; Beyenburg et al., 2009; Szaflarski et al., 2010).

This study aims to compare tolerability, safety and side effect profiles for intravenous and total prophylactic regimens of LEV against PHT, among craniotomy patients requiring intravenous AED, to inform prescribing choice.


Study design

We undertook a pragmatic prospective 3-month open-cohort, single-center study with restricted randomization and appropriate blinding comparing LEV and PHT at standard doses (Fig. 1). We aimed to capture data pertaining to intravenous or early postintravenous administration and to total intravenous-plus-oral-follow-on regimen. Following randomization the study was observational. Primary end points for tolerability were study AED discontinuation because of side effect, and first side effect. Safety end point was major side effect (severity defined below) or seizure. Secondary objectives were comparisons of seizure frequency and side effect severity and quality. Analysis was “as treated” with censoring for known study AED discontinuation, since varying treatment durations were expected and new occurrence of referable side effects after discontinuation unlikely.

Figure 1.

Study design.
AED, antiepileptic drug; IV, intravenous; PO, per oral; bd, twice daily; d, daily. Note minimum of one dose IV AED required, then duration according to treating team (could be >3 days). LEV doses were 250–500 mg bd IV or PO; PHT dose was 300 mg (up to three doses in 24 hours) or 1,000 mg (single dose) IV loading then 300 mg daily IV or PO. *Additional PHT titration to therapeutic serum levels was allowed but not mandated.


Eligible patients were adults >18 years, with neurosurgical indications requiring craniotomy for which perioperative intravenous seizure prophylaxis was routine or otherwise warranted. Participants must have been on no AED or stable dose AED(s) excluding study AEDs for 3 weeks before enrollment, and must not have contraindication to either study medication.


Participants were recruited from a single Australian neurosurgical center practicing routine seizure prophylaxis up to 3 months postcraniotomy. The study was approved in September 2007 by St Vincent’s Hospital Melbourne Human Research Ethics Committee (HREC-D 022/07). Recruitment occurred from May 2008 to January 2010. Primary data collection concluded in April 2010 and secondary source follow-up for missing data in November 2010.


Eligible participants were identified from sequential patients presenting to five neurosurgical treating teams. Participants providing informed consent were randomized.


Treatment with study medications

Up to two oral doses of allocated AEDs were allowed between randomization and intravenous administration. One preoperative intravenous dose was required. Following intravenous AED administration, patients received the same medication orally. Doses were within standard range (see Fig. 1). After randomization, treating teams made all decisions regarding study AED treatment including intravenous and oral durations, serologic monitoring, dose adjustment, and cessation.

Accountability/chain of supply of medication

Study AEDs were supplied by the sponsor (intravenous LEV) and investigators and stored and dispensed by St Vincent’s Hospital investigational pharmacy, which maintained records of AED receipt and distribution. Treating teams were notified of AED allocation via quarantined record-keeper and charted medication to authorize dispensing.


Data recording and side effect survey for each patient was conducted 3 days postoperatively and at discharge from hospital to capture acute and local side effects of intravenous AED, and by telephone 3 months postoperatively to assess total intravenous-plus-oral regimen. Demographic data, concomitant conditions and treatment, and life-threatening and major adverse events were documented. Side effects were queried by open and then directed questioning from a list of 24 potential side effects of the two study AEDs with opportunity to state “other” unlisted events. Cessation of study medication and reason for it, drug interactions and severity, and need for venesection or intravenous catheter change were documented. Data were collected pertaining to subjective mood disturbance using open and then directed questioning. Cognition was subjectively assessed compared with expected by underlying pathology and surgery. Formal neuropsychological testing was not undertaken. Seizure type and number were documented. Participant adherence to medication was estimated by missed doses per week. (See Fig. S1.)

Early exit occurred at next data collection for discontinuation of study AED, or at time of withdrawal for other reason.


Primary outcomes were discontinuation of study AED because of side effect; side effect defined by drug-emergent symptoms determined by the investigators to be definitely, probably, or possibly associated with study AED; and clinically undesirable events defined as major adverse side effects or seizures. Events possibly associated with study medication included, for example, emergent rash or mood lability in respective contexts of concomitant antibiotic and steroid therapy. Events thought very unlikely to be associated with study medication and attributed to other aspects of the patient’s condition or treatment, for example, death from malignancy, were not included as possible side effects.

Major, moderate, and mild side effects are reported. Severity classification took into account nature and outcome of side effects, treating team and patient assessment of severity, documentation by doctors performing questionnaires, and whether there was requirement for study AED cessation. Major side effects comprised life-threatening events or those with significant or potentially significant morbidity requiring AED cessation (including, e.g., rapidly evolving severe rash or arrhythmia, responsive to treatment). Classification was “moderate” for less severe nonprogressive side effects for which study AED was ceased (e.g., mild rash), or where patient or treating team reported a (nonmajor) side effect as moderately severe, and for other clinically moderately severe side effects with or without requirement for AED cessation or dose reduction. Mild side effects did not affect patient function significantly, were self-limiting and/or easy to tolerate, were not reported by the patient or treating team to be more than mild, and did not require AED cessation. Composite major/moderate and mild/moderate outcomes are reported, to assist comparison with other studies.

Seizure occurrence was a secondary outcome. Recruitment target was insufficient to power detection of any small difference in efficacy between LEV and PHT in the prophylactic setting: none was expected.

Study size

Up to 37% of patients administered intravenous PHT experience side effects (Henkin et al., 1996). Our target was 44 patients in each treatment group for power of 0.88 and α = 0.05 to detect a 25% absolute difference in first side effect occurrence for intravenous LEV from an expected 30% event rate for intravenous PHT. Ultimately time constraints limited recruitment to 81, with 74 patients initiating treatment per-protocol, and event numbers were far fewer than anticipated, resulting in lower power.


Block randomization for 88 patients was independently contracted. However, early during data collection the contractor communicated that allocation was as follows: each 10 sequentially recruited patients were not internally randomized but received the same drug, determined by hat-draw at enrollment of the first patient in each block, with eight blocks of 10 patients then two blocks of four to be randomized with equal probability. The allocation procedure was communicated to the quarantined keeper for randomization data, and to an investigator responsible for 90-day data collection, outcome adjudication, and data analysis who remained blinded to allocation. At study completion impact of allocation procedure on bias was assessed by statistical comparison of baseline patient characteristics, with similar age and gender distribution and proportion of serious pathologies and death from underlying pathology found between treatment groups.

Appropriate blinding

The investigation team conducting the information and consenting process, data collection, outcome adjudication, and analysis was blinded. The quarantined keeper and liaison for randomization data did not participate in recruitment, patient treatment, data collection, outcome assessment, or analysis. Recruiting neurosurgical teams including anesthetists were blinded to the allocation procedure. Otherwise the study was open-label.

Data management

Patient data were kept in secure confidential storage. Data sheets including questionnaire responses and medical record data were systematically reviewed and entered into a secure database. Missing data were addressed where possible, according to study protocol and participant consent, by appeal to secondary sources including relatives, carers, and medical records. Secondary source data is acknowledged where applicable.

Statistical methods

Data were characterized initially by descriptive statistics. Percentages (denominator = patients analyzed) were calculated. Fisher’s exact test was used to compare baseline data and post-intravenous data with complete datasets. Survival analysis was planned for comparison of total intravenous-plus-oral AED regimen to allow for censoring to minimize type II error where patients ceased medication before 90 days. Censor time was determined by documented or reported cessation, or where not available, the midpoint between last documented taking of medication and interview time. Survivor functions were presented as Kaplan-Meier curves for the two treatment groups and compared by hazard ratio (HR) with 95% confidence intervals (95% CIs) by the Cox method (Cox, 1972). Zero events in one group for a number of outcomes precluded HRs, hence p-values by the logrank test for comparison of survivor functions were also reported (all p-values were two-tailed). Statistical analyses were conducted using Stata II.2 (StataCorp, College Station, TX, U.S.A.).


Participants and treatment profiles

Patient disposition, demographic and clinical characteristics, and treatment profiles are summarized in Fig. 2 and Table 1. Of 81 patients randomized, 74 (36 LEV, 38 PHT) who received at least one dose of intravenous study AED comprised the population for analysis.

Figure 2.

Patient disposition.
IV, Intravenous; LEV, levetiracetam; PHT, phenytoin; AED, antiepileptic drug; NLR, Study AED no longer required per treating team. *Reasons for premature discontinuation of AED included side effects, rationalization of medication with intercurrent illness and complex medication regimen, charting of alternate medication per usual practice rather than study protocol, palliation, death, and patient preference to pursue alternative therapy. †Secondary source data included information from medical guardians, carers, medical team, and medical records, according to study protocol. This included secondary source information on two patients for whom data sheets were not returned after 3-day data collection. Two patients (both PHT) could not be contacted after discharge with no avenue for secondary source data collection and so were designated lost to follow-up, although one of these is known to have died during the study period from underlying illness and so is included in known deaths (Table 1) and as partial loss to follow-up above.

Table 1.   Patient demographic and clinical characteristics, and treatment profile
Patient characteristics/treatment profileLEV
n (% or median, range)
n (% or median, range)
  1. AED, antiepileptic drug; SDH, subdural hematoma; EDH, extradural hematoma; ICH, intracerebral hemorrhage; SAH, subarachnoid hemorrhage; IHD, ischemic heart disease; HT, hypertension; TIA, transient ischemic attack; DM, diabetes mellitus; EtOH, potentially injurious regular alcohol use; PE, pulmonary embolus; BDZ, benzodiazepine; FET, Fisher’s exact test.

  2. aLEV: Seven glioblastoma multiforme (GBM); one oligodendroglioma/oligoastrocytoma, one anaplastic oligodendroglioma, one olfactory neuroblastoma ethmoid sinus through cribriform plate. PHT: six GBM, one anaplastic astrocytoma.

  3. bLEV: Two melanoma metastatic to brain, one cerebellar metastasis colonic adenocarcinoma. PHT: seven lung, one breast, and one rectosigmoid carcinoma (ca) metastatic to brain; atrial leiomyosarcoma metastatic to brain.

  4. cLEV: Epithelial/myoepithelial ca invading anterior cranial fossa, squamous cell ca ethmoid/anterior cranial fossa and esophageal ca with temporal bone metastatic adenocarcinoma into temporal and midcranial fossae.

  5. dLEV: Includes ICH as complication of metastatic melanoma already counted among secondary brain metastases (the only patient counted twice in primary pathology data).

  6. eLEV: Dermoid cyst posterior third ventricle with hydrocephalus. PHT: Sylvian aqueduct stenosis.

  7. fIncludes baseline and intercurrent comorbidity arising. Comorbidities unlikely to result in complications, such as isolated treated vascular risk factors, are not included. Each patient is counted only once.

  8. gPatients may be counted more than once for different comorbidities.

  9. hLEV: Both patients (one ICH, one GBM) had secondarily generalized seizures prior to enrollment. Neither was on anticonvulsant before enrollment. Neither had seizures during the study. PHT: One patient had generalized tonic–clonic seizures and simple partial seizures, the other had generalized tonic–clonic seizures before enrollment (both had frontal meningioma). Neither was on anticonvulsant before enrollment. Both had seizures in the first week postoperative (Table 3).

  10. iLEV: Melanoma with pulmonary metastases, melanoma, epithelial/myoepithelial ca, olfactory neuroblastoma, squamous cell ca ethmoid, esophageal ca, colonic adenocarcinoma, acute myeloid leukemia in remission (frontal meningioma), history melanoma and breast ca (GBM). PHT: Breast ca with pulmonary metastases, rectosigmoid ca, two non–small cell lung ca one with esophageal mass, lung ca, large cell lung ca, small cell lung ca, clear cell lung ca, pulmonary adenocarcinoma, atrial leiomyosarcoma, previous melanoma (anaplastic astrocytoma),

  11. jLEV: Multiple sclerosis (elective clip aneurysm), hepatitis B on lamivudine (GBM), Crohn’s disease with IHD and strokes (GBM), pleural effusion with IHD, aortic valve repair, atrial fibrillation (AF), pacemaker (meningioma), Sjogren’s syndrome with chronic pancreatitis (meningioma), frontotemporal dementia and Parkinson’s disease with IHD (SDH), previous right cholesteatoma removal (elective aneurysm clip). PHT: Cirrhosis and cytopenia with injurious EtOH use (SDH), myelodysplasia (SDH), chronic obstructive pulmonary disease (COPD) (metastatic lung ca), mitral valve prolapse with permanent pacemaker (atrial leiomyosarcoma), COPD with AF and previous melanoma (anaplastic astrocytoma), Parkinson’s disease with dementia (SDH).

  12. kNone had seizures or major side effects. LEV: Both taking carbamazepine for indications other than seizure prophylaxis. PHT: One taking pregabalin and one taking valproate, both for indications other than seizure prophylaxis.

  13. lNone had seizures or major side effects. PHT: two patients were intubated on propofol; two commenced gabapentin for pain.

  14. mLEV: One patient had midazolam and two temazepam ≤day 3, another temazepam >day3. PHT: Six patients had temazepam, two diazepam and one both ≤day 3. One patient had clonazepam to terminate seizure, three continued diazepam, and one patient each had temazepam, oxazepam (for anxiety) and midazolam >day 3.

  15. nBy unpaired t-test for comparison of means.

  16. oDoes not include one patient lost to follow-up but includes two patients for whom follow-up data were partial.

   FET p < 0.20
N = 74, n (% N)36 (49)38 (51) 
Female18 (50)19 (50) 
Mean age, years58 (56.5, 25–88)60 (60, 27–89) 
90-day data obtained31 (86)30 (79) 
 Intracranial tumour21 (58)20 (52) 
 Primary brain tumora10 (28)7 (18) 
 Secondary brain metastasisb3 (8)10 (26)0.07
 Extraaxial malignancy local intracranial invasionc3 (8)00.11
 Meningioma5 (14)3 (9) 
 Cerebral abscess1 (3)0 
 SDH/EDH5 (14)6 (16) 
 ICH/SAH +/− aneurysm clip4 (11)d6 (16) 
 Aneurysm elective clip5 (14)5 (13) 
 Othere1 (3)1 (3) 
 Open craniotomy32 (89)31 (82)  
 Intervention via burr hole4 (11)7 (18)  
Subjects with serious comorbidityf22 (61)25 (66) 
Baseline comorbiditiesg   
 ≥1 seizure prior to randomizationh2 (5)2 (5) 
 Extraaxial malignancyi9 (25)11 (29) 
 IHD3 (8)4 (11) 
 Arrhythmia2 (5)5 (13) 
 HT15 (42)10 (26) 
 Stroke/TIA1 (3)1 (3) 
 DM4 (11)4 (11) 
 EtOH4 (11)3 (8) 
 Thyroid disease2 (5)1 (3) 
 Migraine/chronic headache3 (8)2 (5) 
 Mood disorder9 (25)4 (11)0.13
 Dementia1 (3)2 (5) 
 Otherj7 (19)6 (16) 
Intercurrent comorbiditiesk   
 Sepsis5 (14)9 (25) 
 PE2 (6)1 (3) 
 New arrhythmia1 (3)0 
Died from underlying illness/complications ≤ 3 months3 (8)5 (13) 
Intercurrent medicationg   
 AED other than BDZ baselinek2 (6)2 (5) 
 AED other than BDZ addedl04 (11) 
 BDZ4 (11)12 (32)0.05
 BDZ ≤ day 3m3 (8)9 (24)0.11
 BDZ > day 3m1 (3)7 (18)0.06
 Opioid21 (58)23 (61)  
 Antibiotic30 (83)30 (79)  
 Dexamethasone22 (61)19 (58) 
 Chemotherapy7 (19)8 (21)  
 Radiotherapy10 (28)14 (37) 
 Antihypertensive (including acute, maintenance)19 (53)19 (50)  
 Antidepressant7 (19)6 (16)  
 Oral hypoglycemic or insulin4 (11)2 (5)  
 Thyroxine2 (5)1 (3)  
Study AED treatment  
 Mean doses IV AED1.39 (1, 1–8)1.34 (1, 1–7)  
 Mean duration treatment AED to discharge days7.7 (6, 1–20)7.7 (6.5, 1–24)  
 Mean duration treatment AED total days64.5 (81, 1–90)54 (54, 1–90)0.19n
 Premature discontinuation other than for side effects5 (14)10 (26)o0.25
 Premature discontinuation all cause9 (25)15 (39)0.22
 Continuation until seizure prophylaxis no longer required27 (75)21 (55)0.09
 Continuation to 90 days19 (53)15 (40)0.35

No significant differences were noted between groups in demographic or other clinical characteristics except BDZ exposure. Nonstatistically significant differences bear mention. Although underlying pathologies were similar, more patients assigned PHT had malignancy metastatic to brain (p = 0.07) and more assigned LEV had extraaxial malignancy with intracranial extension (p = 0.11) (Table 1 and footnotes a–e). Numbers of patients with seizures at baseline and use of other AEDs were similar (Table 1 and footnotes h and k–m). Severity analysis for underlying conditions was beyond the scope of the study; from available data, the number of patients with serious comorbidities was high but did not appear to differ significantly between AED groups (LEV 22, PHT 25, p = 0.81) (Table 1 and footnotes g–j). More patients assigned LEV had baseline mood disorder (p = 0.13) and more assigned PHT took BDZs (p = 0.05), particularly beyond 3 day postcraniotomy (p = 0.06). Of patients taking BDZ, 3/4 LEV and 6/12 PHT took temazepam only.

There was no significant difference between treatment groups in premature discontinuation (LEV 5, PHT 10, p = 0.25) because of palliation/death (LEV 3, PHT 5) or other factors unrelated to study outcomes. Continuation to team determination prophylaxis was no longer required was LEV 75%, PHT 55% (p = 0.09, denominator includes loss to follow up after discharge), with no significant difference between numbers of patients continuing to 90 days (p = 0.35).

Medication rationalization without reported side effect motivated the treating team to stop PHT in one patient, otherwise AED drug interaction was not reported to account for medication cessation. Reported adherence to study AED was near-full except one patient admitted missing oral LEV for up to 3 weeks. Another (PHT) did not receive oral AED in the initial postoperative period before having a seizure (see below).


Outcome data is summarized in Tables 2 and S1A,B.

Table 2.   Outcome data: discontinuation, safety, seizures and side effects for perioperative intravenous and total intravenous plus oral prophylactic AED regimens
n (%)
n (%)
HR LEV versus PHTa95% CI for HRap-Valueb
  1. LEV, levetiracetam; PHT, phenytoin; IV, intravenous; AED, antiepileptic drug; HR, hazard ratio; CI, confidence interval; % values are over number of patients analyzed; n = number of patients experiencing outcome; [n] = total number of outcomes among patients analyzed.

  2. aHRs and CIs by Cox method with accompanying p-value.

  3. bp-values by Fisher’s exact test (FET) for A, and according to the Cox method (Cox-p) where possible and/or the log-rank test for comparison of survivor functions (logrank) for B.

  4. cHR and CI not calculated, as zero events for one comparator.

  5. *Denotes 95% CIs < 1.00 or p-value ≤ 0.05.

A: IV AED: 3 day data plus local complications     FET
Primary outcomes      
 Discontinued because of side effect1 (3)2 (5)1.00
 Any side effect first event [total]1 (3) [1]4 (11) [5]0.36
Other outcomes: side effect profile      
 Major side effect02 (5)0.49
 Local side effect03 (8)0.24
 >1 major or local side effect01 (3)
B: Total IV plus oral regimen to 90 days  CoxCoxCox-pLogrank
Primary outcomes      
 Discontinued because of side effect4 (11)5 (13)0.780.21–2.920.720.71
 Any side effect first event [total]21 (58) [28]14 (37) [17]1.510.77–2.980.220.22
 >1 side effect5 (14)2 (5)
 Major side effect and/or seizure1 (3)10 (26)0.090.01–0.71*0.002*0.004*
 Major side effect and seizure01 (3)
Secondary outcomes      
 Seizure06 (16) c c 0.01*
 Side effect severity      
  Major1 (3)5 (13)0.190.02–1.670.080.096
  Moderate8 (22)4 (11)1.950.59–6.480.260.27
  Mild first event [total]16 (44) [19]8 (21) [8]2.040.87–4.770.090.09
  Major/moderate9 (25)7 (18)1.260.47–3.390.640.64
  >1 major or moderate02 (5)
  Mild/moderate first event [total]20 (56) [27]11 (29) [12]1.850.89–3.870.090.09
  >1 mild or moderate4 (11)1 (3)
C: Exploratory side effect analysis      
 Severe systemic allergy/anaphylaxis02 (5) c c 0.16
 Rash/itch4 (11)5 (13)0.760.20–2.820.680.67
 Allergy and/or major rash04 (11) c c 0.04*
 Drug intoxication02 (5) c c 0.12
 Allergy and/or drug intoxication and/or major rash05 (13) c c 0.02*
 Mood/irritability7 (19)3 (8)2.160.56–8.360.250.25
 Lethargy/tiredness/asthenia8 (22)1 (3)7.710.96–61.650.01*0.02*

No significant difference attributable to intravenous injection was found between LEV and PHT in discontinuation because of side effect (LEV 1/36, PHT 2/38, p = 1.00), or number of patients with at least one side effect (LEV 1/36, PHT 4/38, p = 0.36). No significant difference was found between LEV and PHT total intravenous-plus-oral regimen in discontinuation because of side effect (HR 0.78, 95% CI 0.21–2.92, p = 0.72), or number of patients with at least one side effect (HR 1.51, 95% CI 0.77–2.98, p = 0.22).

Of five patients (one LEV, four PHT) with side effect(s) potentially referable to intravenous AED, three discontinued (one LEV, rash; two PHT, anaphylaxis, severe allergic reaction). Three patients exposed to intravenous PHT developed thrombophlebitis (p = 0.24). No other local side effects were reported. During oral AED follow-on three further patients discontinued LEV because of side effects (one delirium [major] resolving on cessation of medications including LEV, one persistent pruritus and one headache, resolving on discontinuation) and three further patients discontinued PHT because of side effects (one drug intoxication and rash, one moderate, and one mild rash). Another patient reported ataxia and nausea resolving on planned PHT cessation at 90 days; drug intoxication was included as major side effect but not as discontinuation event. Of patients assigned LEV, 21 reported a first side effect compared with 14 assigned PHT (HR 1.51, 95% CI 0.77–2.98; p = 0.22), with side effect totals of 28 and 17, respectively.

In safety analysis, the clinically undesirable combined end point of major side effect and/or seizure occurred in 10 patients assigned PHT and one (delirium) assigned LEV (p = 0.002). Six patients assigned PHT had seizures (p = 0.01; Table 3) including one with undetectable PHT level after seizure day 6. In the remainder, seizures occurred within 4 days of craniotomy. Of the five with serum PHT levels available, only one had a therapeutic level. PHT with relevant dose adjustment was continued in all six patients; one subsequently discontinued because of allergy. None had seizure recurrence to end of study participation. PHT levels during admission were available for 20 (53%) of 38 patients. Five (25% available) had therapeutic levels ≥40 μm (≥10 μg/ml).

Table 3.   Characteristics of patients with postoperative seizure
Patient with seizure (all PHT, chronologic)PathologySeizure time: Days postoperativeSeizure type postoperativeMinimum PHT level (μmol/L) <24 h of seizure
  1. PHT, phenytoin; SDH, subdural hematoma; GBM, glioblastoma multiforme; GTCS, generalized tonic–clonic seizure; CPS, complex partial seizure; SPS, simple partial seizure.

  2. No patients who had postoperative seizures took anticonvulsant before randomization. Patients 7 and 22 took temazepam nocte ≤day 3; Patient 7 was given clonazepam for seizure day 6.

  3. aPatient 7 had multiple comorbidities including syndrome of inappropriate antidiuretic hormone secretion, Na 120 mm.

  4. bSubtherapeutic serum levels (<40 μmol/L). Postanalysis investigation determined Patient 7 received 1 g IV PHT perioperatively with PHT omitted postoperatively until a seizure occurred when PHT IV loading was repeated and oral PHT continued.

  5. cAlso had seizures prior to study enrollment: prior to enrollment patient 22 had GTCS and SPS; Patient 79 had GTCS.

7aChronic SDH6GTCSUndetectableb
22cFrontal meningioma<3SPSNot assessed
27Parietal GBM4GTCS36b
43Temporoparietal GBM3GTCS35b
79cFrontal meningioma3GTCS39b
81Frontal SDH1CPS58

In severity analysis, although not statistically significant, there was a trend for major side effects in more patients assigned PHT (LEV 1, PHT 5, p = 0.08). Four patients with major side effects had metastatic malignancy (one LEV, three PHT), two of those (one LEV, one PHT; melanoma and breast cancer respectively) also had pulmonary metastases (Table 1 footnote i). Mild or combined mild/moderate side effects occurred in more patients assigned LEV (both p = 0.09). Exploratory side effect analysis determined more patients taking PHT were reported to have allergy and/or major rash (p = 0.04) and allergy, drug intoxication, or major rash (p = 0.02). More patients taking LEV reported lethargy/tiredness/asthenia (p = 0.02). New or worsened mood disturbance was reported in seven patients assigned LEV and three assigned PHT (HR 2.16, 95% CI 0.56–8.36, p = 0.25).

Cognitive disturbance deemed out-of-keeping with underlying pathology was specifically reported in four patients assigned LEV and three assigned PHT. This subjective measure was excluded from side effect tallies to avoid double-counting.

Results summary

No significant difference was found in discontinuation because of side effect, or number of patients with side effect, between the two AED groups, either attributable to parenteral AED administration or over the total intravenous-plus-oral regimen to 90 days. More patients assigned PHT had early seizures and/or reached a clinically undesirable combined end point for safety. Local side effects occurred only in association with PHT (not statistically significant). There was trend for major side effects in more patients assigned PHT and for mild side effects in more assigned LEV.


Decisions regarding postcraniotomy seizure prophylaxis are complicated, as seizure rates vary with pathology, and available evidence, largely concerning older AEDs, is heterogeneous (Rossetti & Stupp, 2010) with side effect considerations secondary to efficacy. Side effect burden is highly relevant in determining indication for prophylaxis given relatively large numbers needed-to-treat in postcraniotomy prophylactic settings (Foy et al., 1992; Kuijlen et al., 1996; De Santis et al., 2002; Klimek & Dammers, 2010; Vecht & Wilms, 2010). In this study we found both LEV and PHT to be well-tolerated in intravenous preparation perioperatively, and in total intravenous-plus-oral prophylactic regimen.

Two recent studies have directly compared LEV and PHT. A prospective, randomized, single-blinded trail of intravenous LEV versus intravenous fosPHT (loading)/intravenous PHT among 52 (34 LEV, 18 PHT) critical care patients with severe traumatic brain injury (89%) or subarachnoid hemorrhage (SAH) (Szaflarski et al., 2010), compared neurologic outcome, seizure frequency, and serious adverse events (mean duration intravenous AED 7 days). Significantly fewer gastrointestinal side effects (p = 0.04) occurred among patients assigned LEV, with measures of neurologic status (p = 0.02) and long-term outcome also favoring LEV. No dermatologic complications were reported. Outcomes did not include allergic reactions, total adverse events, or discontinuation because of intravenous AED. Differing side effect profile for PHT in our study (few gastrointestinal side effects, salient occurrence of rash/allergy) may relate to our general craniotomy population and lower mean duration of intravenous AED.

A retrospective efficacy and tolerability study for LEV versus PHT monotherapy after supratentorial neurosurgery (105 LEV, 210 PHT) found a significant difference in number of adverse drug reactions requiring change in therapy during hospitalization favoring LEV (LEV 1%, PHT 18%, p < 0.001), with 64% continuation for LEV against 26% for PHT at 12 months (p = 0.03) (Milligan et al., 2008). The PHT group had greater mean age. Our smaller, shorter prospective study with initial parenteral administration and age comparable between groups found consistent continuation rates, but failed to confirm significant differences in continuation or in discontinuation because of side effect.

Other studies concerning tolerability and safety of parenteral and oral PHT and LEV as individual agents or compared with other AEDs/placebo are heterogenous with differing side effect thresholds, and divided between acute seizure (PHT), chronic epilepsy, neurointensive care, and neurosurgical settings.


Discontinuation of PHT because of side effect has been reported at 5% for intravenous use (Martinelli & Muhlebach, 2003), with between 6.5% (Coplin et al., 2002) and 48% (Henkin et al., 1996) requiring infusion rate reduction, and at 12.6% for rapid oral initiation (Ramsay et al., 2010). Total side effect rate is mostly reported around 25–27% for intravenous or oral PHT (Earnest et al., 1983; Appleton & Gill, 2003; Depondt et al., 2011) but has been cited at 9.1% (intravenous) (Coplin et al., 2002) and 55.9% (oral) (Ramsay et al., 2010). The most common reported intravenous side effects are burning pain at the site, 9.1–37% (Earnest et al., 1983; Henkin et al., 1996; Mattson, 1996; DeToledo & Ramsay, 2000; Coplin et al., 2002; Appleton & Gill, 2003); local cutaneous reaction (LCR)/purple glove syndrome (PGS), 1.3% strict PGS to 25.2% mild to moderate LCR (Burneo et al., 2001; O’Brien et al., 2001); hypotension, 1–14%, higher in SE (Earnest et al., 1983; Binder et al., 1996; Henkin et al., 1996; Mattson, 1996; DeToledo & Ramsay, 2000; De Santis et al., 2002; Appleton & Gill, 2003; Martinelli & Muhlebach, 2003); and drug intoxication, approximately 15% (Earnest et al., 1983). Important but less common side effects are allergy, 2% (rapid loading) (Martinelli & Muhlebach, 2003) and cardiac arrhythmia, 1–7% (Earnest et al., 1983; Mattson, 1996; Treiman et al., 1998; DeToledo & Ramsay, 2000; Appleton & Gill, 2003). Severe morbidity and mortality including tissue necrosis (Twardowschy et al., 2009), requirement for limb amputation (Spengler & Arrowsmith, 1988), Stevens-Johnson syndrome (Delattre et al., 1988), and death from cardiovascular complication (DeToledo & Ramsay, 2000) occur rarely.

Discontinuation of 10% for oral PHT to 1 year after craniotomy is reported (Beenen et al., 1999). Side effect profile of oral PHT mostly relates to drug intoxication. In two recent studies side effect rates were somnolence/sedation 5%, 14.2%; unsteadiness/dizziness/vertigo/ataxia 4%, 17.9%; rash 4%, 7.9%; fatigue 8.7%; abnormal vision 7.1%; and gum hypertrophy 5% (Ramsay et al., 2010; Depondt et al., 2011).

We found consistent discontinuation and total side effect rates for PHT. Rash was most common and most commonly associated with discontinuation in keeping with previous report (Foy et al., 1992, postcraniotomy population PHT/carbamazepine). Concomitant medication, particularly antibiotics, might have inflated rash/allergy rate for both study AEDs. The most severe adverse reactions potentially attributable to PHT were systemic allergic reactions in two patients, including one instance of hypotension, both resolving after discontinuation of medications including PHT. No isolated symptomatic hypotension was reported. Operative monitoring and intervention might have reduced the incidence and reporting of perioperative hypotension, common from other causes. Fewer local side effects than expected following intravenous PHT might relate to perioperative setting with anesthetic technique, sedation, and analgesia. Low PHT serum levels during admission might have resulted in fewer early dose-dependent side effects than otherwise.

Our baseline data showed more patients assigned PHT had brain metastases, although total extraaxial malignancy numbers were comparable between LEV and PHT groups (Table 1 and footnotes b, c and i). Of the seven patients who discontinued or had major side effects or both, four (one LEV, three PHT) had metastatic disease, one bilateral subdural hematoma and alcohol-related liver disease, and one SAH, and one (Patient 79 Table 3) meningioma and seizures (the three last all PHT). As a group, the six patients who had postoperative seizures had less baseline comorbidity (one had ischemic heart disease, myelodysplasia, and hyponatremia; one had atrial fibrillation and previous transient ischemic attack), although one died during the study period with glioblastoma multiforme. We cannot discount that more severe disease, particularly metastatic malignancy in the PHT group, might account for some difference in major side effect outcomes, although baseline disease severity is difficult to compare from our data and more side effects in total were seen among patients assigned LEV (Tables S1A,B summarize outcomes against patient characteristics).

Higher BDZ exposure among patients assigned PHT may have related to more extracranial malignancy and death from primary pathology in this group (not statistically significant). (Six of the 12 patients assigned PHT taking BDZ had extracranial malignancy; two of those, and a further patient with glioblastoma multiforme who took BDZ, died during the study period. Two of the five patients taking PHT with major side effects and another two of the six with seizures, had some BDZ exposure by study end).


Eight studies of intravenous LEV monotherapy or add-on therapy involving populations with brain tumors, SE, in critical care, or with need for oral or PHT substitute (total n = 205, 40 children) have reported infusion well-tolerated with no need for discontinuation because of side effects (Knake et al., 2007; Goraya et al., 2008; Ruegg et al., 2008; Berning et al., 2009; Beyenburg et al., 2009; Moddel et al., 2009; Ng et al., 2010; Usery et al., 2010). Side effects were generally mild, with somnolence, fatigue, nausea, and vomiting most common, and total rates of from very few up to around 30%. A critical review in SE (Trinka & Dobesberger, 2009) composited adverse event rate for intravenous LEV to 7.1%, mostly mild transient side effects. A safety and pharmacokinetic study at high doses and/or infusion rates among healthy subjects (Ramael et al., 2006) found no need for discontinuation, but 86% intravenous LEV versus 25% placebo experienced mild to moderate side effects, prominently dizziness (52.8%), somnolence (33.3%), and fatigue (11.1%). Short-term tolerability of intravenous LEV as oral substitute was evaluated (Baulac et al., 2007), with 20% subjects experiencing related adverse events, all mild or moderate. Two studies specifically reported no local injection site side effects for intravenous LEV (n = 48 combined) (Knake et al., 2007; Berning et al., 2009).

Discontinuation rates because of adverse events for oral LEV have been reported as 19% (long term study n = 811, adverse event/ inefficacy) (Depondt et al., 2005), and 14.4% (28 weeks) (Brodie et al., 2007). Retrospective 2-year retention rate of 53.6% in 196 patients is reported (Chung et al., 2007). Side effect rates for oral LEV monotherapy or add-on therapy have been reported from 6.4% (7-day follow-up) (Zachenhofer et al., 2011) to 88.8% (high dose add-on) (Cereghino et al., 2000), mostly between 30% and 70%, although not differing from placebo in large prospective double-blind multicenter trials (Ben-Menachem et al., 2000; Betts et al., 2000; Cereghino et al., 2000; Shorvon et al., 2000; Bird & Joseph, 2003; Newton et al., 2006, 2007).

Most common side effects for oral LEV include somnolence/sedation and asthenia (14.8% vs. 8.4% placebo and 14.7% vs. 9.1% respectively in review of add-on therapy) (Harden, 2001), dizziness, mood and behavior problems, and thrombocytopenia, with nausea, headache, visual blurring, and rash in a small proportion of patients often not more than placebo (Crepeau & Treiman, 2010). One prospective observational study (LEV add-on, n = 200, ≥6 months) (Bird & Joseph, 2003) found minor adverse effects in 27.4% (corresponding with our mild and moderate side effects), with 16% withdrawal because of major adverse effects, inefficacy, or exacerbation of seizure frequency. Most common reported adverse effects were sleepiness 17.7%, aggression 10.3%, dizziness 4.6%, headache 2.3%, and rash 1.1%.

Our findings correspond with previous discontinuation and total side effect rates and relative frequency of mild to moderate side effects, particularly lethargy/tiredness/asthenia, among LEV-treated patients. LEV-emergent mood-related side effects at upper end of reported range (Briggs & French, 2004) were mild to moderate. Number of affected patients did not differ significantly between LEV and PHT (p = 0.25).

Seizures postcraniotomy

Our finding of no postoperative seizures among patients taking LEV versus PHT 6, p = 0.01, contrasts with previous reports of no significant difference in seizure frequency. In direct comparison: 5/34 LEV versus 3/18 PHT by continuous electroencephalography (cEEG) within 72 h postsurgery, p = 1.0 (Szaflarski et al., 2010); 1/105 LEV versus 9/210 PHT ≤7 days postsurgery, p = 0.17 and 11/42 LEV versus 42/117 PHT at 12 months, p = 0.34 (Milligan et al., 2008). Across postneurosurgical studies: reported seizure occurrence was LEV 2.6% (n = 78, 7 days postoperative) (Zachenhofer et al., 2011), 2/12 (to 4 weeks) (Usery et al., 2010); PHT 7/50 (all with subtherapeutic PHT levels, to 1 year) (Beenen et al., 1999); PHT 13/100 versus placebo 11/100 (7 days, mostly add-on) (De Santis et al., 2002); with no significant difference between PHT, carbamazepine, and no treatment n = 276 (to 6 or 24 months, high proportion subtherapeutic levels) (Foy et al., 1992).

Consistent with previous reports (Beenen et al., 1999; De Santis et al., 2002; Milligan et al., 2008), seizures in our PHT group occurred early postcraniotomy. Generally low available PHT serum levels during admission in our study must be considered in interpretation of comparatively high seizure occurrence in the first week postoperatively, and reflect our real-world setting (cf studies reviewed in Kuijlen et al., 1996). Bias in pathology did not seem to account for seizure outcomes (Tables 1 and 3). Low preoperative seizure occurrence (4/74, 5%; Table 1 and footnote h) in our total study population limit the significance of our findings. Nevertheless our results suggest that LEV compares well against PHT for safety in early seizure prophylaxis after craniotomy.

Choice in intravenous AED

Choice in intravenous AED is needed to manage patient specific factors and to minimize multiple AED exposure. In the neurosurgical setting, LEV’s theoretical advantages over PHT include lack of interaction with chemotherapeutic agents (Yap et al., 2008), dexamethasone (Lawson et al., 1981), and antibiotics, avoidance of worsening cutaneous side effects of radiotherapy (Mamon et al., 1999), and potential neuroprotective benefit (Szaflarski et al., 2010). LEV’s milder side effect profile appeals for neurosurgical populations with high baseline morbidity though mood side effects, lethargy (Wen et al., 2006), thrombocytopenia and cost are potential drawbacks. Our safety data lend support to empirical use of intravenous LEV in critical care (Ruegg et al., 2008; Szaflarski et al., 2010) and for prolonged SE (Knake et al., 2007; Berning et al., 2009; Moddel et al., 2009; Trinka & Dobesberger, 2009). Theoretical basis exists for potential advantage in other groups where PHT may be less suitable including those with intracerebral hemorrhage (ICH) (Naidech et al., 2009), SAH (Naidech et al., 2005), the elderly (Beyenburg et al., 2009), women of childbearing potential, those with hepatic impairment or taking hepatically metabolized drugs (Briggs & French, 2004; Crepeau & Treiman, 2010) including other AEDs and warfarin, and with CYP2C9 variant alleles (Depondt et al., 2011).

The PHT pro-drug fosPHT (intravenous or intramuscular) is not available in Australia. Main theoretical advantages of its water soluble formula, especially pertinent to acute seizure settings, are ability for rapid administration and reduction in local and cardiovascular side effects attributed to the propylene glycol carrier of intravenous PHT (Coplin et al., 2002). Rapid administration is necessary to achieve bioequivalence to intravenous PHT, with salient side effect pruritus responsive to infusion rate reduction (DeToledo & Ramsay, 2000; Coplin et al., 2002). Data from our perioperative setting included few acute local (three instances thrombophlebitis) and no acute cardiovascular side effects for intravenous PHT that may have been ameliorated were fosPHT used instead.

Optimal durations for postcraniotomy seizure prophylaxis are not determined (Temkin et al., 1990; Martinelli & Muhlebach, 2003; Rossetti & Stupp, 2010). However, our safety data support both suggestions that where AED prophylaxis is used consideration should be given to LEV first-line where appropriate (Rossetti & Stupp, 2010; Vecht & Wilms, 2010) and recommendations for close titration of serum levels (Martinelli & Muhlebach, 2003), particularly during the first postoperative week, where PHT is chosen.


Study on a larger sample is required to validate our findings, with power considerations accounting for smaller effect size in the perioperative setting. Necessity to exclude a subset of patients already taking AEDs is likely to have reduced representation of those with low grade and/or highly epileptogenic lesions. Idiosyncratic randomization technique, potentially open to anticipation, introduced greater error vulnerability based on temporal recruitment factors than the planned block randomization. Recruitment from total intake of five different neurosurgical teams was expected to mitigate recruitment and participant outcome reporting bias conferred by the AED allocation method. Blinding was limited as appropriate in our setting.

Compromise of allocation procedure concealment appears not to have introduced significant selection bias (Table 1). Any bias in serious pathology had potential to affect outcomes and censoring analysis. Bias favoring PHT was expected where early discomfort from intravenous AED might have occurred perioperatively or intraoperatively.

We did not direct serum PHT monitoring to optimize therapy in this pragmatic study, meaning seizure and safety results might in part reflect individual pharmacokinetic responses and site-specific management. The low proportion of therapeutic PHT serum levels achieved imply that we did not represent ideal PHT efficacy and capture side effect burden with close titration in the early postoperative period. Perioperative setting and concomitant therapy including analgesia, antibiotics, steroids, chemotherapy, and radiotherapy, were expected to confound side effect reporting. Different pathologies and associated treatment could have confounded side effect profiles and seizure frequency. Conceivably, study AEDs themselves could have modulated patient complaints. We did not undertake independent systematic monitoring for cytopenia or hepatic derangement and did not formally assess cognition, although relevant reports and side effects were included in baseline and side effect data. Monetary cost was not addressed.


Comparative tolerability, safety, and efficacy of intravenous LEV presents broader opportunity for tailoring of drug-to-patient and continuity between intravenous and oral treatment phases. Larger clinical trials are necessary to establish whether alternatives to intravenous PHT might be more effective in prevention of early seizures following craniotomy, and their value in late seizure prophylaxis.


The authors thank the following for contributions to patient enrollment: T. Han, P. McNeill, B. O’Brien, C. Thien, K. Bulluss, C. Chung, J. Peters-Willke, and P. Smith; to data collection: M. Cardamone, A. French, H. Gardner, S. Punchihewa, and N. Shuey; to administration and/or data management: R. Jeffery, A. Roddick, L. Litewka, S. Cook, B. D’Souza, J. Hundal, S. Kalra and the investigational pharmacy at St Vincent’s Hospital; and S. Petty, L. Sedal, S. Vogrin and R. Kapsa for technical assistance. We gratefully acknowledge general statistical advice provided by S Vander Hoorn (University of Melbourne). This was an investigator (MC) initiated study conducted with the assistance of funding and provision of intravenous levetiracetam from UCB Pharma and was conducted independent of other input or review by the sponsor.


Karen Fuller has received an unrelated travel grant from UCB Pharma. Mark Cook has received travel grants and educational honoraria from UCB Pharma, educational honoraria from Sanofi Australia, and travel grants from SciGen Pharmaceuticals. Wendyl D’Souza has received travel, investigator-initiated, and speaker honoraria from UCB Pharma, educational grants from Novartis Pharmaceuticals and Pfizer Pharmaceuticals, educational, travel and fellowship grants from GSK Neurology Australia, and honoraria from SciGen Pharmaceuticals. Yi Yuen Wang and Michael Murphy report no conflicts of interest. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.