Infantile Spasms: Early Treatment May Improve Neurodevelopmental Outcomes

INFANTILE SPASMS: EARLY TREATMENT MAY IMPROVE NEURODEVELOPMENTAL OUTCOMES

Andrew Lux

Introduction

Infantile spasms is the term given to the most common epilepsy syndrome with onset in infancy. It has been the focus of many studies, in part because of the high probability of children with infantile spasms having poor neurodevelopmental outcomes. In considering the possible effects of infantile spasms upon cognition and behaviour, both during the period in which clinical epileptic spasms are occurring and during subsequent life, it is essential to consider several factors and relationships related to their pathophysiology. The most important of these are (1) the nature of the probable underlying cause (or aetiology) for the infantile spasms; (2) the effects that any identified aetiology might have upon the proneness to infantile spasms or other forms of epilepsy, and upon longer term prospects for development; and (3) the causal relationships that are known to exist or might exist between aetiology, the epileptic spasms, and neurodevelopmental outcomes. It is also important to consider questions related to treatment interventions, the key questions being (1) which treatment interventions are most likely to lead to the best cognitive and behavioural outcomes and (2) what is the optimal timing for these interventions.

Classification, terminology and key epidemiological concepts

Because we need to consider the concepts and data gathered from studies over several decades, we also need to appreciate some of the variation and evolution that have occurred in the concepts, classification and terminology relating to infantile spasms. Two issues merit specific emphasis: (1) terms relating to the case definition of infantile spasms and West syndrome and (2) terms relating to aetiological categories.

Since the 1960s, the International League Against Epilepsy (ILAE) has produced several proposals and recommendations on the classification of seizures and epilepsies (Gastaut 1969; Gastaut 1970; Scheffer 2012). These proposals have evolved over time, with substantial additions in the 1980s (Anon 1981; 1989), and the development of a clear five-axis diagnostic framework and glossary in 2001 (Blume et al. 2001; Engel and International League Against Epilepsy 2001). In 2010, there were modifications suggesting a more multidimensional approach to classification and a revised terminology for aetiological categories (Berg et al. 2010; Berg and Scheffer 2011; Scheffer 2012).

CASE DEFINITIONS OF INFANTILE SPASMS AND WEST SYNDROME

It was long considered that arrest or regression of development at the time of onset of infantile spasms was one feature of a triad that constituted the epilepsy syndrome called West syndrome. However, in 2004, a consensus group of 30 clinicians from 15 countries presented the West Delphi consensus statement on case definitions and outcome measures for use in clinical trials of infantile spasms, and stated the argument that, even though regression of development was common and usual in this condition, it was also difficult to assess reliably in young infants and ought not to be a necessary defining feature of the epilepsy syndrome (Lux and Osborne 2004). The group recommended that the term West syndrome is reserved for cases where infantile spasms – that is, the seizure type that can be more generically referred to as epileptic spasms – is associated with hypsarrhythmia.

Hypsarrhythmia is an EEG pattern consisting of a combination of a very high-amplitude and asynchronous background and associated frequent, multifocal epileptiform discharges. It is a pattern that is usually found between, rather than during, seizure attacks, and because of its chaotic nature, it is generally considered to be a manifestation of severely disordered brain function. Although it is a very distinctive EEG finding, it is subject to an appreciable degree of inter-rater unreliability (Hussain et al. 2015). During the epileptic spasms in a particular infant, one or more of several different EEG patterns may occur (Kellaway et al. 1979; Fusco and Vigevano 1993).

Given that the terms infantile spasms and West syndrome have been used inconsistently over the years, and that there are practical problems with reliably identifying EEGs as hypsarrhythmic, this review will use the more generic term infantile spasms to refer to both situations unless otherwise specified.

AETIOLOGICAL CLASSIFICATION OF INFANTILE SPASMS

Studies and clinical reports have also been inconsistent in their use of long-established terms applied in epileptology to categories of aetiology – the terms symptomatic, idiopathic and cryptogenic (Lux and Osborne 2006). More recent ILAE recommendations are that these terms should be replaced by the categories genetic, structural/metabolic, and unknown cause (Berg et al. 2010), but use of these categories can also lead to ambiguity and uncertainty. When analysing and interpreting aetiological data from studies of infantile spasms, however, it is usually possible to map cases with such labels onto the categories of proven aetiology and unknown aetiology, which are the terms that will be used in this review unless otherwise specified. It should be borne in mind that, as diagnostic capabilities improve, we would expect a period effect of proportionately more cases falling into the category of proven aetiology.

CAUSAL PATHWAYS WITH INFANTILE SPASMS AND NEURODEVELOPMENT

In considering the causal relationships between any identified underlying aetiology, infantile spasms, and neurodevelopmental outcomes that include cognitive and behavioural outcomes, we need to consider the factors that might bias, confound, or modify these relationships. Such factors might include the sex of the child, the age at onset of the spasms, the age at which treatment is first started, the nature of the treatment or the educational attainment of the parents.

Some of these factors, such as the effectiveness of any treatment intervention, are of significant interest in the interpretation of clinical studies. Other factors, such as age at onset of spasms and educational attainment of parents, are unlikely to be the prime focus of attention in clinical studies but can be used to assess the robustness of any identified association that might be potentially causal. As such, data about such factors can be used to assess the potential bias, confounding, or effect modification by such variables by including them in an appropriate data analysis. For example, if the variables of interest were thought to contribute to the effects of a measure of cognitive outcome at the age of 5 years, they might be included as a predictor variable in a linear regression analysis with cognitive outcome as the dependent variable, perhaps identifying them as significant independent predictors of cognitive outcome, or adjusting for their effects.

THE INFLUENCE OF AETIOLOGY UPON NEURODEVELOPMENTAL OUTCOMES

There are many underlying causes of infantile spasms. A review of causes of epileptic encephalopathy by a group in Australia has suggested the classification of genetic causes into structural abnormalities, metabolic conditions, recognisable clinical syndromes, and specific genes (Kamien et al. 2012). Non-genetic causes include hypoxic-ischaemic encephalopathy, arterial ischaemic stroke, central nervous system infections, autoimmune conditions, tumours and trauma.

In the United Kingdom Infantile Spasms Study (UKISS), 207 infants were considered for enrolment in the randomised controlled trial (Osborne et al. 2010). Of those infants, 127 (61%) had a proven aetiology, 68 (33%) had unknown aetiology, and 12 (6%) were considered to have been insufficiently investigated to permit reliable categorisation.

Given that many neurological conditions are associated with infantile spasms, and that the incidence of infantile spasms is relatively high, there is an argument that any identified genetic, metabolic, or structural condition that is identified in a specific clinical case might in fact be coincidental and not directly involved in the development of infantile spasms. There is also an argument that, just because one aetiology – Down syndrome, say – has been identified, there might be a good reason to seek other potential aetiologies.

These are reasonable considerations in clinical practice, and posing such questions would help us to avoid the clinical risks associated with cognitive biases such as premature diagnostic closure. However, it is reasonable to assume that, in most cases, there is at least a contribution from any identified aetiology to both the risk of developing infantile spasms and the risk of longer term cognitive or behavioural problems. It is also reasonable to assume that studies presenting information relating to aetiology and neurodevelopmental outcomes are, in general, studying a valid relationship that is amenable to statistical analysis and scientific inference.

Cases with unknown aetiology are important in clinical practice and in the context of trial methodology because it is generally considered that there is greater potential to influence their neurodevelopmental trajectory than with cases that have severe genetic, structural or metabolic causes of infantile spasms. Therefore, it is in the group with unknown aetiology that we are most likely to identify the beneficial effects upon neurodevelopment of novel treatment interventions.

Improving cognitive outcomes following infantile spasms

In order to gain a perspective on our current practice, it is useful to consider the natural history of infantile spasms reported in earlier studies. In the early 1970s, Jeavons et al. (1973) assessed the outcomes for 150 children at the age of 12 years and found that 33 (22%) had died. A study from Nagoya, Japan, during the 1960s and 1970s investigated the outcomes at the age of 6 years or more in 200 children with infantile spasms (Matsumoto et al. 1981). Forty-eight (24%) had died and only 19 (9.5%) were considered to have normal physical and psychological development.

SYSTEMATIC REVIEW OF NEURODEVELOPMENTAL OUTCOMES

The heterogeneity of study design and the focus in earlier studies upon cessation of spasms rather than neurodevelopmental outcomes have led to a relative paucity of data relating to cognitive and behavioural outcomes with infantile spasms. However, investigators in Toronto have made a strong effort to collate data from earlier studies and to provide summary estimates (Widjaja et al. 2015). They aimed to address three main questions: (1) the proportion of cases with good neurodevelopmental outcomes; (2) whether that proportion had changed significantly when comparing studies published before or during 2004 with studies published after 2004 and (3) whether any effects on neurodevelopment might be attributable to differences in lead time to treatment. They also presented a pooled analysis of the proportions with good neurodevelopmental outcome according to whether or not cases had identified aetiologies.

In October 2013, they screened a range of medical databases – Medline, Embase, Cochrane, PsycINFO, Web of Science and Scopus – and identified 1436 studies, either observational or with randomised design elements, of possible relevance. Studies were included in their final analysis if they (1) included five or more participants with infantile spasms; (2) reported outcomes after a median or mean duration of follow-up of more than 6 months and (3) reported neurodevelopmental outcomes. Fifty-five of the articles met those criteria and were analysed further, and all but eight of them had a retrospective study design.

There was no evidence of publication bias (Egger’s test: t = 1.378; p = 0.174). However, there was evidence of significant heterogeneity between studies (Cochran’s Q value = 366.849; p < 0.001), and thus, the pooled estimate was calculated using a random-effects model. The analysis included 2967 patients and the overall proportion with good neurodevelopmental outcome was 23.6% (95% confidence interval [CI]: 19.3% to 28.6%). There were 34 studies published before or during 2004, and 21 studies published after 2004. The proportion with good neurodevelopmental outcome in the earlier study period was 22.0% (95% CI: 16.8% to 28.3%), and in the later study period, it was 26.4% (95% CI: 19.7% to 34.4%). Comparing the estimates from those two periods showed no significant difference in effect size (Q value = 0.862, p = 0.353).

Of the studies included in the systematic review and where aetiological categories were evident, 25 reported outcomes on cases with both proven aetiology and unknown aetiology, 10 studies reported outcomes solely on cases with proven aetiology and 7 studies reported solely on cases with unknown aetiology. Pooled estimates of the proportions with good neurodevelopmental outcome showed significant differences between cases with proven aetiology, where the proportion was 12.5% (95% CI: 9.1% to 17.1%), and cases with unknown aetiology, where the proportion was 54.3% (95% CI: 45.8% to 62.5%) (Q value 69.724, p < 0.001). For the period to the end of the year 2004, a Forrest plot provided a clear illustration of how the proportion with good outcome was consistently higher in four studies where cases had unknown aetiology (Dulac et al. 1993; Vigevano et al. 1993; Gaily et al. 1999; Kivity et al. 2004).

Fig. 11.1 Distribution of Vineland Adaptive Behavior Scale scores at age 14 months in United Kingdom Infantile Spasms Study (UKISS) according to randomly allocated treatment.

Effects of treatment interventions on neurodevelopment

Given that infantile spasms have such a strong relationship with poor neurodevelopment, probably the most important question facing clinicians and parents of a child with infantile spasms is whether there is an available treatment intervention that is likely to have a positive effect upon subsequent cognition and behaviour.

Only gold members can continue reading. Log In or Register to continue