Abstract
Safety and Effectiveness of Hormonal Treatment versus Hormonal Treatment with Vigabatrin for Infantile Spasms (ICISS): A Randomised, Multicentre, Open-Label Trial
Finbar J. K. O'Callaghan, Stuart W. Edwards, Fabienne D. Alber, Eleanor Hancock, Anthony L. Johnson, Colin R. Kennedy, Marcus Likeman, Andrew L. Lux, Mark Mackay, Andrew A. Mallick, Richard W. Newton, Melinda Nolan, Ronit Pressler, Dietz Rating, Bernhard Schmitt, Christopher M. Verity, John P. Osborne; for the participating investigators. Lancet Neurol 2017;16:33–42.
BACKGROUND: Infantile spasms constitutes a severe infantile epilepsy syndrome that is difficult to treat and has a high morbidity. Hormonal therapies or vigabatrin are the most commonly used treatments. We aimed to assess whether combining the treatments would be more effective than hormonal therapy alone. METHODS: In this multicentre, open-label randomised trial, 102 hospitals (Australia [three], Germany [11], New Zealand [two], Switzerland [three], and the UK [83]) enrolled infants who had a clinical diagnosis of infantile spasms and a hypsarrhythmic (or similar) EEG no more than 7 days before enrolment. Participants were randomly assigned (1:1) by a secure website to receive hormonal therapy with vigabatrin or hormonal therapy alone. If parents consented, there was an additional randomisation (1:1) of type of hormonal therapy used (prednisolone or tetracosactide depot). Block randomisation was stratified for hormonal treatment and risk of developmental impairment. Parents and clinicians were not masked to therapy, but investigators assessing electro-clinical outcome were masked to treatment allocation. Minimum doses were prednisolone 10 mg four times a day or intramuscular tetracosactide depot 0·5 mg (40 IU) on alternate days with or without vigabatrin 100 mg/kg per day. The primary outcome was cessation of spasms, which was defined as no witnessed spasms on and between day 14 and day 42 from trial entry, as recorded by parents and carers in a seizure diary. Analysis was by intention to treat. The trial is registered with The International Standard Randomised Controlled Trial Number (ISRCTN), number 54363174, and the European Union Drug Regulating Authorities Clinical Trials (EUDRACT), number 2006-000788-27. FINDINGS: Between March 7, 2007, and May 22, 2014, 766 infants were screened and, of those, 377 were randomly assigned to hormonal therapy with vigabatrin (186) or hormonal therapy alone (191). All 377 infants were assessed for the primary outcome. Between days 14 and 42 inclusive no spasms were witnessed in 133 (72%) of 186 patients on hormonal therapy with vigabatrin compared with 108 (57%) of 191 patients on hormonal therapy alone (difference 15·0%, 95% CI 5·1–24·9, p=0·002). Serious adverse reactions necessitating hospitalisation occurred in 33 infants (16 on hormonal therapy alone and 17 on hormonal therapy with vigabatrin). The most common serious adverse reaction was infection occurring in five infants on hormonal therapy alone and four on hormonal therapy with vigabatrin. There were no deaths attributable to treatment. INTERPRETATION: Hormonal therapy with vigabatrin is significantly more effective at stopping infantile spasms than hormonal therapy alone. The 4 week period of spasm cessation required to achieve a primary clinical response to treatment suggests that the effect seen might be sustained, but this needs to be confirmed at the 18 month follow-up.
Commentary
Infantile spasms (IS) is an epileptic encephalopathy that poses challenges for early diagnosis and effective treatment, monitoring response to treatment, and measuring long-term outcomes. IS varies in clinical presentation; can have atypical EEG findings besides hypsarrhythmia; and lacks diagnostic, treatment, and follow-up guidelines. One consensus exists on the urgency to induce remission defined as cessation of spasms and hypsarrhythmia (1). O'Callaghan et al. of the International Collaborative Infantile Spasms Study (ICISS) must be applauded. This multinational randomized study of 102 centers (83 in the United Kingdom) represents the largest IS trial (n = 377) thus far, and compared corticoids (oral prednisolone or intramuscular tetracosactide depot, n = 191) monotherapy with combined (vigabatrin + oral prednisolone or vigabatrin + intramuscular tetracosactide, n = 186) therapy. Investigators used simple exclusion criteria and a clinically weighted 42-day (treatment begins day 0) protocol. Neither patients nor clinicians could be masked to treatment allocation due to the trial design. In addition, the trial allowed choice of corticoids by parents, and this skewed the number of infants on the oral prednisolone, being 2 to 3 times the number on the intramuscular tetracosactide in each arm. Primary outcome was clinical: ‘cessation of spasms during 14 to 42 days’. A key component of EEG in defining the primary outcome was lacking. IS outcomes are best evaluated by clinical assessments and video-EEG (VEEG). A multihour or overnight VEEG that records multiple awake and sleep stages and transitions and identifies any ictal events is ideal (1–3). Measuring response from a seizure diary without confirmation with the VEEG, as used in this trial, can miss subtle spasms or variants of spasms, mislabel an ictal EEG cluster as an ‘isolated spasm’ or ‘no spasm’ (subclinical) due to electroclinical dissociation, and may label nonepileptic head nods as spasms. This is more likely in infants with severe brain diseases, hypotonia from disease or treatment, and partial responders. Investigators attempted to study the electroclinical response as a secondary outcome by combining a blinded review of EEG done between days 14 and 21. However, EEG protocol at the trial entry and on treatment was unclear, and how the EEG duration, recording of awake–sleep stages and transitions, and capturing of ictal events impacted the trial is unknown. At trial entry, 55 patients were not treatment naïve and were on concurrent antiepileptic treatment (unspecified) of ‘other seizure types’. The ratio of screened to enrolled patients was 50%, which is not uncommon in such trials. However, there was no documentation of reasons for excluding 50% of screened patients. In theory, a selection bias in enrollment cannot be excluded.
The primary outcome measure showed a difference between groups, with 72% spasm cessation in the combined versus 57% in corticoids group. The response rate in the corticoids monotherapy group was lower than previously reported and generally accepted. A previous trial by the same investigators reported a much higher responder rate of 73% from corticoids monotherapy, but the study was under powered and measured outcome of 48-hours spasm cessation (4). As a secondary outcome, when EEG was added to determine electroclinical response, response rates dropped to 66% in combined versus 55% in corticoids groups. Responder rates dropped steeply in the combined (72%–66%) but modestly in the corticoids (57%–55%) groups, which is intriguing and emphasizes difficulties in labeling IS remission on the basis of parental reports that can be optimistic. On days 13 to 14, 89% in combined and 69% in corticoids group had cessation of spasms (P < 0.001) but 57 (19%, 33 combined and 24 corticoids group) relapsed by day 42. The trial allowed local investigators to increase doses on any day between days 7 and 14 for nonresponders/early relapsers but then assessed primary outcome beginning day 14. If some patients had treatment escalation just before or on day 14, they may not have had enough time to reach the steady pharmacological state on the higher dose of the long acting/depot drugs and could have been classified as nonresponders for the primary outcome. The trial suggested that the chance of achieving an electroclinical response with oral prednisolone was less than with intramuscular tetracosactide, but due to the trial design, this result should be interpreted with caution. Past studies have reported similar findings, however, the evidence remains insufficient (5).
IS studies have shown that short-term remission, occurrence of relapse, and long-term epilepsy and cognitive outcome strongly depend on the etiology. Systematic documentation of refined etiology is lacking in the trial despite all centers being in developed countries. Investigators simply separated the subjects into those with or without risk factors of development impairment by using clinical ‘risk factors of development impairment’ as a surrogate marker. Of the 58% of the cohort who had an etiologic diagnosis, most were in the severe brain disease group. It is no surprise that children at ‘high risk of developmental impairment’ showed poor primary outcome and electroclinical response rates compared with the ‘low-risk’ group in both treatment arms.
The trial noted that greater than 2 months of delay in initiation of treatment was associated with lower rate of cessation of spasms, emphasizing the need to urgently recognize and treat IS. The authors do not report significant differences in the adverse events in the two groups. However, in Table 5, O'Callaghan et al. show sedation in 24% (4 significant) in the combined compared with 2% in corticoids group. Sedation may reflect vigabatrin side effect or exaggerated pharmacodynamic/pharmacokinetic interaction of vigabatrin with corticoids and/or other antiepileptic drugs (unspecified) and should be a caution especially in children with severe brain disease, hypotonia, multiple organ-system involvement, and risk of aspiration or infection.
Investigators plan to follow the cohort and report 18 months outcome, which will be of even greater value to know relapse rates, emergence of other catastrophic epilepsies, and cognitive development in combined versus corticoids group. There is a suggestion from a previous monotherapy trial that children with cryptogenic IS (low risk for developmental disabilities) who were allocated to corticoids had a better neurodevelopmental outcome at 14 months and 4 years (6, 7) compared to those who received vigabatrin. Whether corticoids and vigabatrin combination therapy used in the cryptogenic group would offer even further improvement in long-term outcomes compared to corticoids monotherapy has not been determined, but would be an interesting question to have answered to further strengthen arguments for early combination therapy in the cryptogenic group.
This ICISS trial highlights aggressive and early treatment of IS. Combined therapy may offer advantages in infants who are relatively healthy and lack ingredients of severe brain disease but it must be done with close monitoring for side effects, objectively documenting response via VEEG, and assessing benefits and risks at periodic intervals. In infants with severe brain disease, serial therapy may still be a better alternative due to risk of side effects, guarded prognosis for developmental outcome over the long term, and cost. In IS due to an epileptogenic brain MRI lesion highly associated with intractability and amenable to surgical resection, benefits and risks of combined trials, and other treatments should be carefully weighed against delaying surgery. Further trials with rigorous assessment of electroclinical outcomes as well as systematic capture of etiology need be done to establish if combined therapy as a standard of treatment for all IS or a subset of IS can be accepted.