This week sees the publication of two hugely anticipated randomised controlled trials both published in The Lancet simultaneously which DFTB have been given exclusive access.
ConSEPT was conducted by the PREDICT research group
Dalziel S, Borland M, Furyk J, Bonisch M, Neutze J, Doriath Sm Francis K, Sharpe C, Harvey AS, Davidson A, Craig S, Phillips Nm George S, Rao A, Cheng N, Zhang M, Kochar A, Brabyn C, Oakley E, Babl F. Levetiracetam versus phenytoin for second-line treatment of convulsive status epilepticus in children (ConSEPT): an open-label, multicentre, randomised controlled trial. The Lancet, https://dx.doi.org/10.1016/S0140-6736(19)30722-6
EcLiPSE led by Prof. Richard Appleton in collaboration with PERUKI
Lyttle M, Rainford N, Gamble C, Messahel S, Humphreys A, Hickey H, Woolfall K, Roper L, Noblet J, Lee E, Potter S, Tate P, Iyer A, Evans V, Appleton R, with support of Paediatric Emergency Research in the United Kingdom & Ireland (PERUKI) collaborative* Levetiracetam versus phenytoin for second-line treatment of paediatric convulsive status epilepticus (EcLiPSE): a multicentre, open-label, randomised trial. The Lancet https://dx.doi.org/10.1016/S0140-6736(19)30724-X
Why were these studies needed?
Convulsive status epilepticus is a common paediatric emergency with significant consequences for the patients. Our focus on the management of status is to stop the seizures quickly to avoid any complications which range from requiring an RSI to neurological sequelae (34%) and occasionally death (3-5%). All guidelines use benzodiazepines as the first-line treatment with significant evidence to support this. However only 40-60% of patients in status have their seizures terminated with benzodiazepines, so second-line treatments are crucial.
Most guidelines recommend phenytoin (or fosphenytoin in the US) as a second-line treatment but much less comparative work has been undertaken. Phenytoin is linked to many adverse events including liver damage, Steven-Johnson syndrome, extravasation and reports of death due to dosing errors. As a result, and because of its biopharmacology, it is a resource-intensive drug to make up in an emergency.
Levetiracetam is an alternative to phenytoin for second-line treatment of convulsive status epilepticus. It can be given over 5 minutes (phenytoin takes 20 minutes to infuse), is more compatible with IV fluids, has fewer drug interactions, and has a lower risk of adverse events.
Although small studies suggest that levetiracetam is equally effective, there have been no comparison studies…until now.
The aim of these studies was to establish whether levetiracetam or phenytoin is better as a second-line treatment for the management of convulsive status epilepticus in children.
Who were the patients?
These were children (3 months to 16 years) presenting to one of 13 emergency departments in Australia and New Zealand over a 2.5 year period.
To be eligible, children had to present in convulsive status epilepticus and have received two doses of benzodiazepines (either by parents, pre-hospital staff, or hospital staff). There were minimum dosing requirements for eligibility although the actual dose was as per the local hospital guidelines.
Convulsive status epilepticus was defined using the International League Against Epilepsy definition:
– unresponsive with ongoing abnormal movements or increased tone for longer than five minutes
– two or more recurrent seizure episodes with no recovery of consciousness in between
– three or more convulsions within the hour and currently convulsing
Exclusions were due to being on regular phenytoin or levetiracetam; having already had a second-line treatment; known to be refractory to the medication; allergy; or seizures secondary to a head injury.
639 patients were eligible for enrolment. After exclusions, 114 were in the phenytoin group, and 119 were in the levetiracetam group (based on intention to treat).
EcLiPSE recruited from a slightly older start date (6 months) but recruited up to 18 years of age across 30 Emergency Departments in just under 3 years.
Patients were included if the treating clinician felt the patient was in status. Exclusion criteria were very similar to ConSEPT. A deferred consent approach was used whereby the clinicians would randomise without parental or carer involvement and obtain consent following the study drug infusion. A separate study was embedded into EcLiPSE to examine this consent process.
The treatments were decided by a pre-defined randomisation schedule (1:1). Treatment packs only being opened when the patient was confirmed to be eligible after the final first-line treatment had been administered. As the effect time of benzodiazepines is variable, and either treatment takes time to be drawn up, some study participants would have stopped fitting by the time the drug was ready to be administered.
After assessing 1432 patients for eligibility 404 were randomly allocated to treatment, 192 to phenytoin and 212 to levetiracetam of which, on an intention to treat basis, 134 in the phenytoin and 152 in the levetiracetam arm received treatment.
For both studies, the objectives of the study were clear and relevant. The patient group is well defined and due to the nature of status, it is likely that were very few incorrect inclusions. All subjects were accounted for in both groups and the deferred consent process appears to have worked well in both studies. The differences in age selection may have been significant for the younger group in relation to case identification (i.e seizure activity in less than 6 months old can be difficult to determine) but both groups had relatively similar proportions of ages.
Patients were randomised 1:1 and whilst a double-blind study would have been optimal, we realise that this was not possible. Both studies were analysed on an intention to treat analysis with some patients being randomised but then promptly intubated prior to having the study drug.
What was the intervention?
Patients were randomised to either phenytoin or levetiracetam. All of the team and the parents knew which group they had been allocated to as masking was not possible.
The phenytoin group received 20mg/kg (max 1g) phenytoin IV or IO over 20 minutes
The levetiracetam group received 40mg/kg (max 3g) levetiracetam IV or IO over 5 minutes.
The EcLiPSE study used the same loading doses but slightly different maximums from ConSEPT
The phenytoin group received 20mg/kg (max 2g) phenytoin IV or IO over 20 minutes
The levetiracetam group received 40mg/kg (max 2.5g) levetiracetam IV or IO over 5 minutes.
What outcomes were measured?
Five minutes after the infusion finished, patients were assessed to see if the seizures had stopped (determined by the presence of increased tone or jerking movements).
The primary outcome was seizure cessation 5 minutes after the drug infusion and where possible the seizure cessation was verified independently via a video recording to reduce observer bias between the two groups.
If the patient was still convulsing, then the other drug was given (i.e. if the patient had levetiracetam they would now be given phenytoin). If the patient was still seizing 5 minutes after that infusion then they were managed according to the local guidelines (which was usually RSI and intubation).
Secondary outcomes were: time to cessation of seizures; the need for other medications/interventions; adverse events; ICU admissions; airways complications; arrhythmias; length of hospital stay; or seizure status up to two months later.
In a key difference from the ConSEPT study, the primary outcome was time “from randomisation to cessation of all visible signs of convulsive activity, defined as cessation of all continuous rhythmic clonic activity, as judged by the treating clinician”. As per the inclusion criteria, this is a very real-world pragmatic approach.
Secondary outcomes were essentially the same as ConSEPT with the addition of the list of serious adverse reactions (which include death, Stevens-Johnson syndrome, rash, airway complications, cardiovascular instability, extravasation injury, and extreme agitation, as well as those listed in the summary product characteristic of each treatment).
The ConSEPT authors acknowledged the challenge with the timing differences because of the infusion rates. As the phenytoin infusion takes 20 minutes, patients in that group were assessed 25 minutes after the start of the infusion; levetiracetam infusion only takes 5 minutes, and so patients in that group were assessed 10 minutes after the start of the infusion. A similar situation occurred in EcLiPSE, although not so pronounced because of the active cessation of seizure definition, but technically the phenytoin group had a greater time to recover.
It can be difficult in practice to determine when a seizure has stopped, and the ConSEPT authors recognised this possibility of observer bias in the study design by video recording the episodes and having them reviewed independently. The EcLiPSE study took a more pragmatic approach, which may have led to potential inaccuracies in determining definitive seizure cessation but this also reflects everyday practice more closely. Conversely, the knowledge that you might have your decision making recorded may have altered clinical practice in the ConSEPT group.
What did the results show?
The sample was powered for the hypothesis that levetiracetam is superior to phenytoin (this required 91 patients in each group).
Although the study was based on intention to treat, some patients were excluded due to needing RSI/intubation before the medication was given, therefore a modified intention to treat analysis was performed too.
The groups were well matched for demographics and baseline characteristics.
- Mean age was 3.9 years
- 26% had developmental delay
- 47% had previous seizures
- 24% had previous convulsive status epilepticus
Interestingly the length of seizure was 73 minutes (median) before patients received the first study drug.
The primary outcome – cessation of seizures 5 minutes post the infusion – occurred in 60% in the phenytoin group and 50% in the levetiracetam group.
After 2 hours, 54% of the phenytoin group and 51% of the levetiracetam group did not require further medications and remained seizure-free. 37% of patients in the phenytoin group received levetiracetam, and 40% of patients in the levetiracetam group received phenytoin as the seizures had not stopped. 18% in the phenytoin group v 26% in the levetiracetam group required intubation.
It took 22 minutes to cease seizure activity from commencing phenytoin and 17 minutes from commencing levetiracetam (median time).
The data is well-described and the groups were comparable at baseline. The results are clearly presented in sufficient detail to allow readers to interpret and make their own assessment. The study was powered to assess whether levetiracetam was superior to phenytoin which is another key difference from the EcLiPSE (which required more numbers but was designed to show a difference between the drugs if there was one there – see below)
The aim of the EcLiPSE was to determine which was the more effective and safer second-line agent with the methodology section describing a sample size designed to detect a difference in seizure cessation rates between the two drugs from the previously cited rate of 60% (in phenytoin) to 75% (in levetiracetam). The original calculation determined that 308 patients would be needed to allow for a 10% loss to follow up ultimately providing 140 participants per group. At the interim analysis, it was found that there was very low attrition and so the final sample was adjusted to 286.
From a pool of 1432 screened patients i.e those presenting with a seizure 404 were randomly allocated to a study medication and final numbers, accounting for seizure cessation prior to drug delivery and failure to gain consent, were 134 in the phenytoin group and 152 in the levetiracetam group. Three patients in the levetiracetam group actually received phenytoin but were analysed in this group on an intention to treat basis. Similar to ConSEPT, in EcLiPSE there was equal matching of age ranges and pre-morbid conditions between groups.
The median time to seizure cessation was 45 mins in the phenytoin group (IQR 24 to not assessable) and 35 mins in the levetiracetam group (IQR 20 to not assessable due to RSI). Participants in both groups received additional anticonvulsants (37% with phenytoin and 38% with levetiracetam) and only five serious adverse events were observed in total split between both drugs, four being short-lived and one a child dying secondary to cerebral oedema which was felt to be unrelated to either treatment. There we no other significant differences in relation to PICU admission or RSI use.
As per ConSEPT all patients are accounted for and there is a huge amount of detail on the analyses used to correct for patients who had additional drugs or didn’t have an outcome time for seizure termination as they were intubated. The EcLiPSE study is available open access so this information is available easily to all.
The authors noted they used a superiority design as they had expected levetiracetam to be more effective and safer. There was no reason to believe on the previous evidence that this wouldn’t be the case so their study design is reasonable.
What did the authors conclude?
Levetiracetam is not superior to phenytoin as a second-line agent for convulsive status epilepticus
There is no significant difference between phenytoin and levetiracetam in the second-line treatment of paediatric convulsive status epilepticus for any outcome, including time to seizure cessation
Should we change practice?
Two studies were released simultaneously asking an identical question – it’s a field day for evidence on status epilepticus in children! Depending on your viewpoint or biases it may be sad to see no definitive conclusion from either study. The hope had been that levetiracetam would be proven to be a more effective agent and that this would drive forward changes in APLS algorithms.
While there were differences between the study designs, the primary outcome measure of timing being the largest, the fact that both studies found no difference probably means head-to-head there is little difference.
Given that EcLiPSE should have had a tendency for shorter seizure cessation times (as these were recorded as soon as the seizure stopped) that this may have aided ConSEPT (which had a defined time point for review) in demonstrating a difference, but this wasn’t born out.
The nature of the statistical analysis means that both groups rightly point out that in their cohorts levetiracetam wasn’t superior in outcomes to phenytoin. A future pooled analysis could still demonstrate a difference, but it seems unlikely that a critical difference will be seen (especially for the safety element).
Given the wealth of evidence on the side effects of phenytoin, it is surprising the incident rates were relatively low. Whether in study conditions more care was taken with drawing up and delivering the drug or that previous safety reviews were heterogeneous in their inclusion criteria is difficult to know. However, the time to draw up phenytoin, and the background concerns on its potential harm, will lead some to suggest that the switch to levetiracetam is a logical one, regardless of its effectiveness against phenytoin.
The challenge faced by many units is the capacity for PICU beds. Because phenytoin is given over 20 minutes there is time to prepare for airway/anaesthetic intervention if it is unsuccessful in terminating the seizure. The use of levetiracetam may cause some to wonder if they should then try phenytoin either as a stop-gap to bed availability or because the time in status now seems ‘shorter’ than normal. These are not statistical issues, these are pragmatic clinical conundrums.
The absence of a clear winner will further fuel this debate meaning it is unlikely in the immediate future we are going to see a change from the ALSG or similar organisations. However, local units may decide, in the clear absence of harm from levetiracetam, that it is a drug they should be adding to their treatment protocols.
Post-publication commentary from the lead authors
Fantastic commentary Damian and Tessa. Given the risk and resources associated with RSI and intubation, it makes logical sense that algorithms will actually change over time. In fact, both New Zealand’s and Victoria’s algorithms for status epilepticus are in the process of changing to Benzodiazepine x 2, then levetiracetam, then phenytoin, then RSI. The main reason for this change is that such a guideline halves the number of children being recommended RSI by the guideline compared to a guideline using just levetiracetam or phenytoin alone, at the expense of only an additional 10 minutes; halves the number of children exposed to phenytoin, including the risk for rare serious adverse events; and in the 20 min that the phenytoin is being given (as you comment) the team can prepare for RSI, previous research has shown that it takes time to do this and that we have not been as good as we would like to be in the past at meeting time goals for seizure management.
Stuart Dalziel (lead author ConSEPT)
Damian and Tessa have written a really clear, concise and informative synopsis of these two important, if not pivotal studies. ’EcLiPSE’s and ’ConSEPT’s findings demand digestion and debate. The results of both studies might well lead to the possible conclusion (perhaps some would argue more appropriately, temptation), to give both second-line drugs sequentially to potentially avoid RSI, its implications and potential complications. Whilst this may seem attractive from one angle, caution must be observed because not all children with convulsive status epilepticus are the same; there are often different aetiologies and different durations of status. A potential – or real – delay in RSI and ventilatory support may be detrimental to the child with an increased risk of morbidity, if not mortality. Wisdom should dictate a multi-speciality forum comprised of paediatric emergency medicine, neurology, intensive care, general paediatrics and pharmacy. What happens in the Emergency Department does not always stay in the Emergency Department…
Richard Appleton (chief investigator EcLiPSE)
For those of you who enjoy a visual summary, here is an infographic by @gracie_leo
Of course, we love doing them.
If there are any particular papers you think we should tackle then just let us know.
I really enjoyed reading this study that has been broken down into PICO and critically appraised. Please could you offer more such studies as I am terrible at critical appraisal and i find it really helps to read articles like these (that have a critical appraisal attached) so that I can learn to do something similar! Thanks.