•
`
`Seizure 23 (2014) 167-174
`
`Contents lists available at ScienceDirect
`
`Seizure
`
`ELSEVI ER
`
`journal homepage: www.elsevier.com/locate/yseiz
`
`Review
`The relative effectiveness of five antiepileptic drugs in treatment
`of benzodiazepine-resistant convulsive status epilepticus:
`A meta-analysis of published studies
`
`1) CrossMark
`
`Zeid Yasirya'*, Simon D. Shorvon b
`'Department of Medicine, University of Babylon/College of Medicine, Babil, Iraq
`b Department of Clinical and Experimental Epilepsy, UCL Institute of Neurology, London, UK
`
`ARTICLE INFO
`
`ABSTRACT
`
`Article history:
`Received 30 November 2013
`Received in revised form 14 December 2013
`Accepted 16 December 2013
`
`Keywords:
`Status epilepticus
`Meta-analysis
`Lacosamide
`Levetiracetam
`Phenobarbital
`Phenytoin
`Valproate
`
`Purpose: Systematic evaluation of published evidence-base of the efficacy of five antiepileptic drugs -
`lacosamide, levetiracetam, valproate, phenytoin and phenobarbital - in convulsive benzodiazepine-
`resistant status epilepticus.
`Methods: Data sources included electronic databases, personal communication, and back tracing of
`references in pertinent studies. These were prospective and retrospective human studies presenting
`original data for participants with convulsive benzodiazepine-resistant status epilepticus. Interventions
`were intravenous lacosamide, levetiracetam, phenobarbital, phenytoin and valproate. Outcome
`measured is clinically detectable cessation of seizure activity. Level-of-evidence was assessed according
`to Oxford Centre of Evidence-Based Medicine and The Cochrane Collaboration's Tool for Assessment of
`Risk. Twenty seven studies (798 cases of convulsive status epilepticus) were identified and 22 included
`in a meta-analysis. Random-effects analysis of dichotomous outcome of a single group estimate
`(proportion), with inverse variance weighting, was implemented. Several sources of clinical and
`methodological heterogeneity were identified.
`Results: Efficacy of levetiracetam was 68.5% (95% CI: 56.2-78.7%), phenobarbital 73.6% (95% CI: 58.3-
`84.8%), phenytoin 50.2% (95% CI: 34.2-66.1%) and valproate 75.7% (95% CI: 63.7-84.8%). Lacosamide
`studies were excluded from the meta-analysis due to insufficient data.
`Conclusion: Valproate, levetiracetam and phenobarbital can all be used as first line therapy in
`benzodiazepine-resistant status epilepticus. The evidence does not support the first-line use of
`phenytoin. There is not enough evidence to support the routine use of lacosamide. Randomized
`controlled trials are urgently needed.
`© 2013 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
`
`I. Introduction
`
`Status epilepticus (SE) is a neurological emergency with
`significant morbidity and mortalityL2 and has to be treated in a
`timely manner before irreversible neuronal damage ensues."
`Having a protocol for therapy is universally recommended, and
`standard protocols are widely accepted.' All of these recommend
`benzodiazepines as first line therapy7" 7 and there is now global
`consensus on this. In contrast, what action to take if benzodia-
`zepines are ineffective is much less clear and there is perceived to
`be a lack of evidence to support the use of any particular agent
`currently employed in the protocols. Because of this paucity of
`evidence, this review was conducted with the aim of examining,
`
`critically, the evidence relating to the efficacy of five anti-epileptic
`drugs in the treatment of benzodiazepine-resistant status epilep-
`ticus. These medications are lacosamide, levetiracetam, valproate,
`phenytoin and phenobarbital. The last two drugs have been
`extensively used for this indication for many years, based largely
`on the evidence derived from the Veterans Affair Trials; although it
`is worth noting that these medications were sometimes given as a
`first-line treatment in that study. The other three antiepileptic
`drugs have been more recently introduced, and although widely
`prescribed in this situation, are not licensed specifically for use in
`status epilepticus.
`
`2. Methods
`
`2.1. Aims
`
`* Corresponding author at: Department of Internal Medicine, College of
`Medicine/University of Babylon, Babil, Iraq. Tel.: +964 (0)7707117766.
`E-mail address: z.yasiryeucl.ac.uk (Z. Yasiry).
`
`To identify, via reproducible methodology, all the available
`literature related to the use of the five anti-epileptic drugs in
`
`1059-1311/$ - see front matter © 2013 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
`http:Alx.cloi.org(10.1016(.i.seizure.2013.12.007
`
`ARGENTUM Exhibit 1063
` Argentum Pharmaceuticals LLC v. Research Corporation Technologies, Inc.
`IPR2016-00204
`
`EXHIBIT
`
`1
`45
`19Z
`d -JD --it‘'7741
`
`Page 00001
`
`

`
`168 (cid:9)
`
`Z Yasiry, S.D. Shorvon/ Seizure 23 (2014) 167-174
`
`benzodiazepine-resistant status epilepticus, to assess the hetero-
`geneity and reliability of the data, to analyze the extracted data to
`quantify the relative efficacy of these drugs, and to provide
`recommendations for the use of the latter in patients with
`benzodiazepine-resistant status epilepticus.
`
`2.2. Patients, methods and analysis
`
`A pre-specified protocol was followed for the search, extraction,
`and analysis of data following the methodology of the "Systematic
`Reviews: Centre of Review and Dissemination's guidance for
`undertaking reviews in health care" published by the Centre of
`Review and Dissemination, University of York' s and "Cochrane
`Handbook of Systematic Reviews of Intervention".19 Patients
`reported in the published papers were included in the analysis
`if they fulfilled the eligibility criteria set out in Table 1. All patients
`with convulsive status epilepticus, of any type, and who had failed
`to respond to benzodiazepine therapy and were thus given one of
`the five study drugs as second-line therapy were included,
`regardless of age or other clinical variable.
`Internet-based searches were implemented through the online
`databases MEDLINE and EMBASE, both accessed via Ovid (see
`supplementary material 1 for search protocol). The search results
`from the two databases were combined with the duplicates
`excluded. In addition, the references in the bibliographies of the
`relevant papers were individually searched and back-traced. In
`several instances, the authors of the identified studies were
`contacted via email or telephone, to answer specific queries
`relating to data analysis in their papers (notably to ascertain details
`of such aspects as the numbers of patients treated who were
`benzodiazepine-resistant and their outcome).
`The papers were selected for the review by screening the search
`results by title and abstract for eligibility. The filtered studies
`would, then, be read as a whole, subjected to the inclusion criteria,
`stratified according to the intervention of interest, and scrutinized
`for their level of evidence and risk of bias. Then, they would go
`
`Table 1
`Eligibility criteria.
`Participants (cid:9)
`
`Interventions (cid:9)
`
`Patients with status epilepticus who have been resistant to
`initial therapy with benzodiazepines were included. Only
`human studies and studies of convulsive (motor) status
`epilepticus were included. In some studies, simple and
`complex partial seizures were not subdivided, and it is thus
`possible that some non-convulsive cases were included:
`however where a study exclusively included non-
`convulsive status epilepticus, it was not considered. There
`was no restriction by age groups, co-morbidities or epilepsy
`background.
`
`Intravenous lacosamide, levetiracetam, valproate,
`phenytoin, and phenobarbital as second line therapy after
`failure of benzodiazepines. No dose or rate restrictions were
`specified.
`
`Comparators (cid:9)
`
`None
`
`Outcomes (cid:9)
`
`Study design (cid:9)
`
`The variable extracted was cessation of seizure activity
`(other outcomes were also sought but are not reported here
`including, mortality, new neurological deficit, and
`tolerability). Cessation of seizure activity, or the drug's
`efficacy, was defined differently by different authors in the
`selected papers, and definition was, therefore, reported as a
`variable and acknowledged as one of the several sources of
`heterogeneity.
`
`Original papers with any study design were included. There
`was no restriction on the number of patients in case series.
`All studies which provided data on outcome following
`treatment with one (or more) of the five drugs were
`included, whether these were controlled or uncontrolled
`and whether or not a comparator was included.
`
`through data extraction, tabulation, pooling then meta-analysis, if
`eligible for the latter.
`Papers were excluded where original data was not presented
`(for example reviews and expert opinions), which were published
`in non-English languages without abstract/accredited translation
`for the required data, where the drugs were used in more advanced
`stages of status epilepticus (where benzodiazepines, then anaes-
`thetics and other antiepileptic drugs had been used before the
`medications of interest), and where data extraction/interpretation
`was not possible.
`The papers were classified into levels according to the Oxford
`Centre of Evidence Based Medicine (CEBM).2° In case of random-
`ized trials and non-randomized prospective studies, assessment of
`the risk of bias was performed using the Cochrane Collaboration's
`Too! for Assessment of Risk."
`Data was extracted by filling out a proforma by one reviewer;
`the process was supervised by the other reviewer. Data were then
`analyzed using both STATA® 11 (by StataCorp LP, Texas, USA) and
`Comprehensive MetaAnalysis version 2 (CMA2g-by Biostat®, New
`Jersey, USA). The protocol is based on dichotomous outcome
`analysis of a single group estimate: inverse variance weighting is
`performed for each estimate, followed by random-effects analysis
`of the pooled estimates of all the studies describing an interven-
`tion, taking in consideration both the within-study and between-
`studies variances. The protocol and formulae for the random effect
`meta-analysis are given in the supplementary material 2. Single-
`patient case reports were not included in the meta-analysis due to
`lack of statistical dispersion. There was one case of epilepsia
`partialis continua found in the review, but as it was a single-patient
`report, it was not included in the meta-analysis.
`The reasons for choosing random-effects model are varying
`sample sources, demographics, aetiology, and types of seizures,
`treatment with different doses, timing of administration, and
`definitions of outcome. All the aforementioned differences are
`substantial sources of heterogeneity that make fixed-effect meta-
`analysis unsuitable. The random-effects model was not chosen
`based on a statistical heterogeneity test.22 However, heterogeneity
`was quantified via 12, a statistic used to quantify how much of the
`variability in the results is due to real heterogeneity rather than a
`random sampling error.23
`
`3. Results
`
`3.1. Characteristics of publications analyzed
`
`A total of 2754 papers were identified on MEDLINE/EMBASE
`(see supplementary material 1) from which 2652 papers were
`excluded due to non-relevance. From the remaining 102 (with an
`added 6 papers from reference tracing), only 27 papers were
`retrieved for data extraction. Some studies covered two or three
`drugs; therefore, the number of papers from summation of studies
`per drug was 32. The papers included consist of 1 randomized
`double-blinded trial, 5 open-label trials, 18 case series and 3 case
`reports. They described 798 episodes of convulsive status
`epilepticus.
`The levels of evidence of the studies are as follows: level 4 (18
`studies, 66%), level 4- (3 studies, 11%), level 2b (5 studies, 19%), and,
`level lb (1 study, 4%) (see supplementary material 3). For
`prospective studies, assessment of the risk of bias was also
`performed, the results of which are illustrated in Table 2. It is worth
`noting that neither the prospective studies nor the single
`randomized controlled trial are registered at the NIH Clinical Trial
`Centre (http://clinicaltrials.govict2Thome).
`Sources of heterogeneity were multiple; these include study
`design (retrospective, prospective, randomized and non-random-
`ized, blinded and non-blinded), demographics (age, gender,
`
`Page 00002
`
`

`
`Z. Yasiry, S.D. Shorvon/ Seizure
`
`23 (2014) 167-174
`
`169
`
`Study name
`
`Event rate and 95% a
`
`Aiguabella et al .,2011
`Alvarez et al., 2011
`Beming et al., 2009
`Eue et al., 2011
`Knale et al., 2008
`Misa et al., 2011
`Ruegg et al., 2008
`Standith et al., 2010
`Summary estimate
`
`Event (cid:9) Lower (cid:9) Upper
`limit
`rate (cid:9)
`limit (cid:9)
`0.500
`0.390
`0.604
`0.432
`0.549
`0.376
`0.403
`0.552
`0.562
`
`0.889
`0.517
`0.818
`0.556
`0.846
`0.700
`0.632
`0.955
`0.685
`
`0.985
`0.642
`0.930
`0.673
`0.961
`0.900
`0.813
`0.997
`0.787
`
`Relative
`weight
`
`5.29
`22.32
`12.97
`22.65
`8.61
`9.96
`15.09
`3.11
`
`1.00
`
`0.50
`
`-0.50 (cid:9)
`
`0.00
`
`-1.00 (cid:9)
`Fig. 1. Forest plot for efficacy of levetiracetam; CI: confidence interval.
`
`comorbidities, and previous medications), intervention character-
`istics (dosage, rate of infusion, manufacture, drug levels), and
`condition characteristics (aetiology, semiology of seizures, dura-
`tion of seizures to be considered status epilepticus, duration of
`status before intervention), response characteristics (time to
`seizure termination, presence of follow up period for re-emerging
`seizures).
`The definition of status epilepticus varied between studies: 10
`papers (37%) used 5-min duration, while 5 other studies (18.51%)
`specified the classical 30-min definition. Ten- and 20-min
`durations of status were the criteria for 2 papers (3.7%, each);
`while 15-min minimum was the criterion for 2 other studies
`(7.4%). In 8 studies (29.6%), a definition for status epilepticus was
`not specified. The definition of response to the intervention varied
`as well. 14 papers (51.9%) specified a time-window in which
`seizure termination was considered favourable. The most
`common specification was termination of seizures within
`30 min of infusion (6 papers, 22.2%); other definitions include
`3 min, 15 min, 20 min, 1 h, 12 h (1 paper for each, 3.7%), 24 h (2
`papers, 7.4%) and 48 h (1 paper, 3.7%). A variable period of seizure
`freedom was a secondary endpoint in 9 studies. The most common
`time-window was 24 h (5 papers, 18.5%); other specified
`windows include 6 h, 12 h, 48 h and 7 days (1 paper for each,
`3.7%). No temporal definition of response was given in 12 papers
`(44.4%). One study (3.7%) linked the time condition for seizure
`freedom to the end of infusion.
`Considering the above mentioned sources of heterogeneity, 12
`was relatively low and within acceptable limits. The raw data from
`the publications included in the analysis are available in the
`supplementary material 4. Because of the heterogeneity or absence
`of data on variables such as age, time of administration, prior
`epilepsy, concurrent AEDs and AED levels, data was not stratified
`according to these variables, although in any future study
`(particularly in a randomized controlled trial) these would be
`important variables to consider.
`
`3.2. Findings
`
`3.2.1. Lacosamide
`After applying the search methods, 109 papers were identified,
`from which only 13 were retrieved, due to non-relevance of the
`rest. From these 13, only 2 papers met the inclusion criteria. The
`papers described treatment of a total of 70 patients with status
`epilepticus of varying aetiologies, semiologies and stages.242s The
`authors provided further data indicating which patients met our
`inclusion criteria (i.e. second-line treatment after benzodiazepine
`failure). Only 4 patients met these criteria, a number too small to
`permit meta-analysis (see supplementary material 4 for details).
`
`3.2.2. levetiracetam
`From original 345 papers identified from the search, 318 were
`excluded by title/abstract screening. While 4 studies were added
`via bibliography tracing, 21 studies were excluded after reviewing
`the whole article. Thus, only 10 papers contributed to this review,
`addressing the use of levetiracetam in 206 SE episodes.26-46 Two
`reports were excluded from meta-analysis because each reported
`only a single patient.2"1 The mean efficacy from the remaining 8
`studies was 68.5% (95% Cl: 56.2-78.7%; Fig. 1). Heterogeneity
`assessed by /2 was 12%. Averaged weighting of each contributing
`study is available on the forest plot as percentage. Two papers,
`those of Eue et al. (2011)26 and Alvarez et al. (2011)36, contribute
`the most to these statistical results.
`
`3.2.3. Phenobarbital
`From 537 search results, 520 studies were excluded via the
`title/abstract screening due to non-relevance. Seventeen papers
`were retrieved of which 3 papers, reporting treatment of 43
`episodes of benzodiazepine-resistant status epilepticus, were
`considered eligible for inclusion.47 -39 One case report was
`excluded from the meta-analysis.' The Meta-analysis revealed
`a mean efficacy of 73.6% (95% CI: 58.3-84.8%; Fig. 2)32 was 0% due
`
`Table 2
`Assessment of the risk of bias in prospective studies.
`
`Study name
`
`Selection bias: random
`sequence generation
`
`Selection bias:
`allocation
`concealment
`
`Performance bias:
`blinding (masking)
`
`Detection bias:
`blinding of outcome
`assessment
`
`Attrition bias:
`incomplete outcome
`data
`
`Reporting bias:
`selective outcome
`reporting
`
`Agarwal et al. (2007)
`Chen et al. (2011)
`Kokwaro et al. (2003)
`Misra et al. (2011)
`Ogutu et al. (2003)
`Malamiri et al. (2012)
`
`Unclear
`Low
`High
`Low
`Unclear
`Low
`
`Unclear
`High
`High
`Unclear
`Unclear
`Low
`
`Unclear
`High
`High
`Unclear
`Unclear
`Low
`
`Low
`Low
`High
`Low
`Low
`Low
`
`Unclear
`Low
`Low
`Unclear
`Low
`Low
`
`High
`Low
`Low
`Unclear
`Low
`Low
`
`Page 00003
`
`(cid:9)
`(cid:9)
`(cid:9)
`(cid:9)
`

`
`170
`
`Z. Yasity, S.D. Shorvon/Seizure 23 (2014) 167-174
`
`Study name
`
`Event rate and 95% CI
`
`Event Lower Upper
`rate
`limit
`limit (cid:9)
`
`Kokw aro et a1.,2003
`Malamiri et al., 2012
`Summary estimate
`
`0.667
`0.767
`0.736
`
`0.376
`0.585
`0.583
`
`0.869
`0.884
`0.848
`
`Relative
`weight
`
`33.20
`66.80
`
`-.4•110w
`
`Fig. 2. Forest plot for efficacy of phenobarbital; CI: confidence interval.
`
`-1.00 (cid:9)
`
`-0.50 (cid:9)
`
`0.00 (cid:9)
`
`0.50 (cid:9)
`
`1.00
`
`Study name
`
`Event rate and 95% CI
`
`Agaves et al., 2009
`Alvarezet al., 2011
`Brevoord et al., 2005
`Franzoni et al., 2006
`Ismail et at, 2012
`Myahara et al.,2009
`Ogutu et al., 2003
`Tiamkao& SaAanyawisuth, 2009
`Summary estimate
`
`Event Lower
`rate (cid:9)
`limit
`0.840
`0.711
`0.586
`0.468
`0.297
`0.198
`0.455
`0.265
`0.265
`0.144
`0.929
`0.423
`0.364
`0.143
`0.459
`0.308
`0.502
`0.342
`
`Upper
`limit
`
`0.918
`0.695
`0.419
`0.659
`0.435
`0.993
`0.661
`0.619
`0.661
`
`-11I-
`
`.111
`
`Relative
`weight
`1a71
`15.37
`15.05
`13.17
`13.67
`4.02
`10.60
`14.40
`
`-1.00 (cid:9)
`
`-0.50 (cid:9)
`
`0.00
`
`-M-
`
`-.01111111--
`0.50 (cid:9)
`
`1.00
`
`Fig. 3. Forest plot for efficacy of phenytoin; CI: confidence interval.
`
`to the number of studies taken, rendering Q statistic = 1 (see
`supplementary material 2 for complete reference of the random-
`effects model). Averaged weighting of each contributing study is
`available on the forest plot as percentage, with Malamiri et al.
`(2012)38 contributing to more than two thirds of the statistical
`weight.
`
`3.2.4. Phenytoin
`There were 996 papers as the result of the protocol used for
`databases search. 968 papers were excluded via title/abstract
`screening. The remaining 28 papers were retrieved for further
`inspection. Only 8 studies, reporting_ 294 episodes of status
`epilepticus, meet the inclusion criteria.36.46-46 Meta-analysis of the
`pooled effect sizes showed a mean efficacy of 50.2% (95% CI: 43.2-
`66.1%; Fig. 3). Heterogeneity via 12 was calculated to be 16.45%.
`Averaged weighting of each contributing study is available on the
`forest plot. Alvarez et al. (2011)'`' and Brevoord et al. (2005)4' seem
`to contribute the most to the statistical results.
`
`3.2.5. Valproate
`After applying the search protocol, 767 results were identified.
`Seven hundred forty two papers were excluded due to non-
`relevance by title/abstract screening; 2 were added via reference
`tracing to give a net total of 27 papers. These were assessed, and
`
`finally 9 papers, describing treatment in 251 benzodiazepine-
`resistant episodes, were included.3b '8 46-6 One case report was
`excluded from the meta-analysis61 while the remaining 8 studies
`yielded a mean effect size for the efficacy of valproate of 75.7% (95%
`CI: 63.7-84.8%; Fig. 4). Heterogeneity calculated via 12 was 12.73%.
`Averaged weighting of each contributing study is available on the
`forest plot as percentage. Alvarez et al. (2011)36 and Chen et al.
`(2011)60 seem to contribute the most to the statistical results.
`
`4. Discussion
`
`4.1. Limitations
`
`To the best of our knowledge, this is the first attempt, to review
`the five antiepileptic drugs for use in patients with status
`epilepticus who have failed to respond to initial benzodiazepine
`treatment (as recommended by most of the current protocols), and
`to implement a meta-analysis of the findings.
`The strength of the study is its strictly applied inclusion criteria,
`and the systematic search, method and analysis. However, the
`investigation revealed a number of important limitations:
`
`a. The number of studies that have addressed the effectiveness of
`second-line therapy is small (27 papers).
`
`Study name
`
`Event rate and 95% CI
`
`Pgarvel et a., 2009
`Alvarez et a., 2011
`Chang et al., 2010
`Chen et al., 2011
`IVItiamiii et a., 2012
`assn et al., 2007
`Tiamkao 8 SaAenyanisuth, 2009
`Yu et al., 2033
`Summary estimate
`
`Event Lower Upper
`rate (cid:9)
`limit (cid:9)
`limit
`
`0.880
`0.746
`0.588
`0.500
`0.930
`0.730
`0.750
`0.969
`0.757
`
`0.758
`0.620
`0.352
`0.328
`0.732
`0.567
`0.448
`0.650
`0.637
`
`0.945
`0.841
`0.793
`0.672
`0.967
`0.848
`0.917
`0.995
`0.848
`
`Relative
`weight
`
`-s-
`-o-
`
`•
`
`-a-
`
`•
`
`14.22
`17.00
`no8
`15.65
`10.97
`15.55
`10.04
`3.50
`
`Fig. 4. Forest plot for efficacy of valproate: CI: confidence interval.
`
`-1.00 (cid:9)
`
`-0.50 (cid:9)
`
`0.00
`
`0.50 (cid:9)
`
`1.00
`
`Page 00004
`
`(cid:9)
`(cid:9)
`

`
`Z. Yasiry, S.D. Shorvon/ Seizure 23 (2014) 167-174 (cid:9)
`
`171
`
`b. The studies are mostly observational and retrospective (21;
`77.7%).
`c. There is near-total absence of randomized double-blinded trials
`(Class I evidence) to compare the interventions (1; 3.7%).
`d. For the few prospective studies, most are open-label (5; 18.5%),
`with high risk of bias in multiple domains.
`e. There is also a lack of homogeneity in the findings - few studies
`sharing comparable questions, outlines of methodology, aims,
`or even definitions of variables (such as status epilepticus) and
`main endpoints (such as the response or its duration). This has
`caused significant clinical and statistical heterogeneity, and
`limited an original intention to study the influence of variables
`in correlation analysis. In addition, this heterogeneity compro-
`mises the strength of evidence derived from this review, as the
`confounders cannot be taken, statistically, into account.
`f. The random-effects model adopted for the meta-analysis has
`resulted in estimates with wide confidence intervals (i.e. larger
`uncertainty) and, therefore, less powerful impact.
`g. In some settings, current practice is to use second-line
`treatment immediately or very soon after a full dose of
`benzodiazepine, especially diazepam, to avoid the potential of
`recurrence of seizures. Neither this practice nor the adequacy of
`first line treatment was appropriately assessed in the partici-
`pating studies.
`
`These limitations are important. Any decisions about drug
`therapy need to be made in the knowledge that the published
`literature is not wholly adequate and that the evidence base on
`which to make comparisons of studies of different designs, with
`different definitions and which do not consider other variables is
`poor. Such a situation though is not uncommon in other clinical
`settings, particularly in relation to emergency therapy. Certainly,
`better quality studies are needed before gold-standard recom-
`mendations can be made. One main outcome of our analysis is to
`highlight these weaknesses. Nevertheless, advice regarding
`therapy is needed, even in the absence of optimal data.
`
`4.2. Choice of drugs
`
`Here we present our findings from the meta-analysis and
`narrative findings regarding side-effects from the published
`literature for each drug. In making comparisons, in the absence
`of any randomized controlled trial (RCT) in which direct
`comparisons are reported, our conclusions must be inevitably to
`an extent subjective. Furthermore, as emphasized in the sections
`on limitations, there are other important clinical factors which
`influence outcome in status epilepticus. Our recommendations are
`therefore provisional and indirectly based, but made on what we
`consider the best available evidence. We have focused on adult SE
`and there are suggestions for some drugs that efficacy in paediatric
`populations may differ. Cost-benefit assessments would also be
`useful, but cannot be made as controlled data related to
`comparison of side-effects and complication rates, and other
`economic variables are not systematically reported. One particular
`issue of interest would be whether treatment, especially in
`refractory cases, may sometimes worsen outcome. This requires
`separate and specific study.
`
`4.2.1. Lacosamide
`There is not enough evidence to recommend using lacosamide
`routinely in the treatment of benzodiazepine-resistant status
`epilepticus at present, despite the accumulating studies com-
`mending on its efficacy in individual cases. Nevertheless, the drug
`has favourable properties including a possible novel mode of
`action and an absence of significant side-effects and interactions,
`which might favour its use in the future especially for patients with
`
`co-morbidities and those on polytherapy.52 From these studies, it
`is clear that Lacosamide can be effective and safe, in a 200-400 mg
`bolus dose range in adult patients; however, the magnitude of this
`efficacy cannot, yet, be compared to the other medications. Data on
`its usage and dosing in paediatric SE is lacking.
`
`4.2.2. Levetiracetam
`The estimated mean efficacy of levetiracetam is 68.5%, when
`infused in doses between 1000 and 3000 mg in young adults, or
`20 mg/kg. Experience is relatively limited, but suggests to date that
`the drug is free of significant adverse-effect and well tolerated in
`paediatric, adult and elderly populations and in those with
`comorbidities. It has neither common cardio-respiratory side-
`effects nor drug-drug interactions.53
`
`4.2.3. Phenobarbital
`Phenobarbital has an estimated efficacy, in the meta-analysis,
`of 73.6%; however, the confidence interval was very wide (95% Cl:
`58.3-84.8%), making the clinical relevance of this result unclear.
`This efficacy, when supported by a potential neuroprotective
`effect, is a significant advantage. Disadvantages include adverse
`effects that limit its use, such as respiratory depression, hypoten-
`sion, severe sedation, tolerance and the potential for drug
`interactions.'54
`
`4.2.4. Phenytoin
`Phenytoin had a mean efficacy estimate of 50.2%. In the
`reported studies, phenytoin was administered in doses classically
`recommended to produce a therapeutic blood level, but it is well-
`established that drug level monitoring is needed in view of the
`non-linear kinetics of phenytoin.''' This was, often, not reported
`(87.5% of studies did not report the levels), and one possible reason
`for relatively low efficacy reported in some patients may have been
`inadequate levels. Another possible explanation is the fact that
`lower cerebral concentrations of phenytoin in animal models are
`found in lesional brain foci of seizure activity.58' Other
`disadvantages are the absence of data substantiating its use for
`older population (due to expected high rate of cardiovascular
`adverse-events) and for neuroprotection, where it may also be
`detrimental in certain types of brain injury.'2 Advantages, though,
`are its long duration of action, fast CNS entry,63'5 availability and
`large experience accrued over decades of use. The side-effects of IV
`phenytoin include significant cardio-respiratory risks (cardiac
`arrhythmia, hypotension, reduced cardiac output)`'`' e" and also
`risks of thrombosis and inflammation at the injection site
`sometimes resulting in distal ischaemia (the 'purple-glove'
`syndrome).' In view of the above points, although phenytoin
`is often considered the drug of first choice in benzodiazepine-
`resistant status, the published evidence does not appear to support
`this practice.
`
`4.2.5. Valproate
`The meta-analysis found the mean efficacy of valproate to be
`75.7%. The fact all the comparative, prospective and randomized
`studies include valproate as one of their two or three arms gives
`more power to the statistical analysis. In addition to its high
`efficacy in acute situation, follow-up seizure freedom rates were
`also higher, and the drug was well-tolerated, even with large doses
`( -up to 100 mg/kg) and rates of infusion (up to -6 mg/kg/min). It
`is free of cardio-respiratory side effects which is an important
`advantage. However, high doses of IV valproate are likely to cause
`hyperammonaemia and in susceptible patients, it is likely that
`ammonia concentrations could rise to very high and potentially
`dangerous levels although data on this is lacking.727 ' There is a risk
`of hepatic and pancreatic toxicity, and valproate encephalopa-
`thy.i3 There is also a theoretical risk that the use of high dose
`
`Page 00005
`
`

`
`172 (cid:9)
`
`Z. Yasiry, S.D. Shorvon /Seizure 23 (2014) 167-174
`
`valproate will exacerbate a bleeding tendency due to its effects on
`platelets and platelet function74, which might carry risks in some
`situations in status epilepticus (for instance in acute stroke), but to
`the best of our knowledge no such side-effects have been reported
`in practice in status epilepticus.
`
`4.3. Other factors influencing outcome
`
`Several significant factors may influence the chance of seizure
`cessation and final outcome(s) of status epilepticus. These have
`been outlined as sources of heterogeneity above, and include
`variables such as the adequacy of first line therapy, the duration of
`status epilepticus before treatment was initiated, the aetiology of
`seizures, the age of the patients, and dosage and rate of infusion of
`the drug. This meta-analysis could not analyze these variables or
`use them to interpret the outcome differences because of missing
`data.
`Several other observations were made about outcome from this
`analysis of the literature. First, there seems wide agreement that
`the duration of seizures before treatment was inversely related to
`the probability of clinical seizure cessation, whatever treatment is
`chosen, i.e. the longer the seizures prior to treatment, the less likely
`they are controlled by medications.40 42 43,46.45 50 Second, aetiology
`is a most important variable to account for both treatment failure
`and adverse long-term outcome, with symptomatic seizures
`carrying the poorest response to second-line treatment.3(436'43•5°
`Finally, there are a variety of findings about whether seizure type
`influences outcome in the various studies of drug effect, and no
`agreement on this point."' 33 " 48 These data require further
`research via prospective designs of future studies in which
`multiple variables are controlled to explore the effects of
`individual factors on primary and secondary treatment outcomes.
`
`4.4. Recommendations for treatment
`
`Although there are few controlled studies and a general poverty
`of data, we consider that the data reviewed above provide
`sufficient evidence for limited recommendations to be made.
`The highest efficacy was attributed to valproate, levetiracetam and
`phenobarbital. Each of these drugs has differing advantages and
`drawbacks. Taken altogether, it is our view that any of these three
`could have claim to be first line therapy in benzodiazepine-
`resistant status epilepticus in most patients. There are differing
`clinical situations where one might be preferred over the others -
`for instance the avoidance of phenobarbital where the risk of
`hypotension or respiratory depression are significant or the
`avoidance of valproate where there is a particular susceptibility
`to hyperammonaemia or hepatic failure (possibly in childr