Clinical manifestations

Presentation and course

Infantile epileptic spasms syndrome is an epileptic encephalopathy that constitutes a distinct and often catastrophic form of epilepsy of early infancy (132). The onset can be abrupt or insidious in either an otherwise normal or an infant with disabilities. It is an age-dependent epilepsy syndrome that begins in infancy, mostly between 4 and 6 months of life, before the age of 12 months in over 90% of cases (90). However, later occurrence, up to 4 years of age, has been emphasized; it is easily overlooked and, therefore, inappropriately treated for many months before the diagnosis is done (14).

Infantile spasms are characterized by usually synchronous, bilateral, brief, and sudden contractions of the axial muscle groups. The features of the seizures depend on whether the flexor or extensor muscles are predominantly affected and also on the number and distribution of the muscle groups involved. Thus, spasms may vary from extensive contractions of all flexor or extensor muscles to contractions of only neck muscles or abdominal recti muscles (75). Spasms may be flexor, extensor, and mixed flexor-extensor. Mixed spasms are most common, followed by flexor spasms, with extensor spasms being the least common (90). Most infants have more than one of these types, and the type observed at any given moment may be influenced by body positions.

The flexor spasm, though not the most common, is the most characteristic. When the abdominal flexor muscles are involved, the body may bend at the waist like a jackknife (jackknife seizure). When the upper extremities are involved, either abduction or adduction of the arms in a self-hugging motion will appear. The jackknife seizure plus the adduction of the upper extremities is reminiscent of the ritual of salaam, thus, the term "salaam attacks". When only the neck flexor muscles are involved, the spasm may be a head nod. The involvement of the shoulder girdle may manifest as a shrug-like movement (90).

Other manifestations have also been described. A behavioral arrest may occur as a seizure without associated spasm. Spasms can be restricted to brief, vertical, ocular deviation or nystagmoid movements. There may be atonic elements combined or uncombined with epileptic spasms (181). Alteration in respiration is also a common associated phenomenon, whereas change in heart rate is rare (90).

The type of spasms does not seem to be affected by etiology or prognosis. In contrast, whether the spasms are symmetrical or not is important because asymmetry may indicate focal cortical brain lesion (59). Asymmetrical spasms may include lateral deviation of the head or eyes. Based on the recent classification of seizures, the possibility that spasms may have a focal onset should be considered.

Spasms tend to occur soon after awakening or on falling asleep. Most of the spasms occur in clusters, ie, the interval between successive spasms is less than 60 seconds. Usually the intensity of spasms in a given cluster will peak gradually and then decline (75). The frequency of spasms varies from only a few times a day to several hundred a day (90). They do not show a predilection for either day or night, although they appear to be temporally related to sleep. Sudden loud noises or tactile stimulation, but not photic stimulation, may precipitate them.

Following a spasm there may be periods of attenuated responsiveness. Crying may frequently follow a spasm, but this is not an ictal phenomenon. In walking children, drop attacks may be the first manifestation of the disorder.

Although spasms in clusters are the hallmark of the syndrome, other kinds of seizures may occur, either clonic or focal. A tonic seizure may initiate a cluster of spasms and a combination of a cluster of spasms with a focal seizure may constitute a single seizure.

The classic interictal EEG is hypsarrhythmia. It consists of diffuse, high-amplitude, nonsynchronous paroxysmal and slow-wave theta and delta activity, with loss of background features. This activity may occur continuously or in bursts and may be present in the waking state, appearing only in non-rapid eye movement (non-REM) sleep and disappearing in REM sleep (06; 172; 180). This hypsarrhythmic pattern is a dynamic condition in which the spikes alternate randomly between focal, multifocal, and generalized discharges at different moments within a brief record (180). Kim and colleagues attempted to determine the brain dynamics during hypsarrhythmia by analyzing the event-related spectral perturbation of fast oscillations (FO-ESRP) on routine EEGs of patients with infantile spasms (93). They found that the locations of high FO-ERSPs are closely associated with brain pathological lesions and can be useful as a biomarker for the localization of pathologies that may not be apparent. Hypsarrhythmia, however, is usually seen in the early stages of infantile spasms, most often in younger infants, and is present in approximately 66% of the cases (21). However, in certain situations, the interictal EEG of patients with infantile spasms may not necessarily be hypsarrhythmic (46). A form of late-onset epileptic spasms with absence of the typical hypsarrhythmic pattern in EEG has been described, and usually in children with known underlying brain pathologies (53; 83). Variations in the classic hypsarrhythmia pattern include increased interhemispheric synchronization; asymmetry; a consistent focus of abnormal discharge; episodes of attenuation; high-voltage bilaterally asynchronous slow activity; excessive rapidity; excessive slowing; fragmentation; and increased periodicity (78). The “chaotic” pattern becomes more organized with time (78; 172) and, between 2 and 4 years of age, may evolve into the generalized slow sharp and slow-wave pattern of Lennox-Gastaut syndrome.

Several ictal EEG patterns are associated with infantile spasms (90). Three are the most common: a high-voltage positive vertex slow wave; a spindle-like fast activity of medium amplitude; a diffuse flattening (decremental response) (166). There is no correlation between the ictal pattern and the type of spasm. The duration of the ictal EEG ranges from 0.5 to 106 seconds. The longer episodes are associated with behavioral arrest. In some instances, the ictal discharge combines focal discharge with the cluster of spasms (24), the focal discharge either preceding, following, or being in the middle of the cluster of spasms. This combination strongly indicates either brain malformation or focal brain lesion. A study investigating the temporal and spatial characteristics of ictal gamma and beta activity on scalp EEG during spasms found that prior to the ictal motor manifestation, focal ictal gamma or beta activity emerged from one hemisphere or from the midline, and was rarely simultaneously bilateral, despite being visually symmetric semiologically (121).

Prognosis and complications

The prognosis for infantile epileptic spasms syndrome in terms of normal development is poor in spite of treatment. Moderate or severe learning difficulties may be present in 70% to 90% of patients at follow-up (141; 142; 144; 160). The prognosis is better in cases that have no known associated etiologic factor, no abnormality on neurologic examination, normal development before the onset of the spasm, and normal neuroimaging prior to therapy (144). A study investigated acoustic auditory processing in patients with infantile spasms of unknown etiology after a 1-month remission and found poor processing of complex acoustic information; this would impair language (56). In a small population-based study in Iceland, approximately one third of children were diagnosed with autism spectrum disorders (146). In children with tuberous sclerosis complex, the occurrence of infantile spasms is a risk factor for autism spectrum disorders (68).

A report from the United Kingdom Infantile Spasms Study (UKISS) shows evidence that increasing lead time to treatment is significantly associated with decreasing developmental score (using Vineland Adaptive Behaviour Scales) at 4 years of age in all infants with infantile spasms after adjustment for the effects of age at onset of spasms, etiology, and treatment allocation. The same study also suggests that age of onset of spasms is important in that earlier onset of spasms is followed by poorer developmental outcome at 4 years of age, independent of the effect of lead time to treatment, treatment allocation, or etiologic group (124). These data are also supported by another study from the U.S. consensus report, which states that regardless of the chosen medication, timely assessment of treatment efficacy (ie, 2 to 4 weeks for ACTH followed by taper; 2 weeks or less following dose titration for vigabatrin) and, if indicated, prompt treatment modification is strongly recommended. This is because longer treatment trials (ie, greater than 2 to 4 weeks for ACTH; greater than 3 months for vigabatrin) are not likely to be effective and may come at the expense of serious adverse events (132).

Patients with infantile spasms carry an increased risk of mortality due to the underlying etiologic disease and comorbid conditions. Riikonen reported that 31% of a Finnish cohort died during a 20- to 35-year follow-up period (142; 143). Trevathan and colleagues examined infantile spasms in an Atlanta cohort (1975-1977) and found that 15% died by the age of 11 and 35% died by the age of 25 (160).

Another center found a 17% mortality during childhood; persistence of epileptic spasms and comorbid respiratory system disorders were the most important determinants of mortality, with early deaths related to neurologic impairments and comorbidities (70).

Clinical vignette

A full-term boy was delivered by cesarean section for breech presentation after normal pregnancy. His parents were not related, and no familial history of neurologic disease could be documented. At two months of age, the parents noticed repeated movements of the upper limbs associated with a blank stare occurring several times a day, especially when the baby awoke.

Video-EEG monitoring with electromyogram (EMG) on deltoids showed interictal background disorganization with high-amplitude slow waves and random multifocal spikes during waking and sleep distinctive of hypsarrhythmia. A predominant anterior left focus was identified. Clusters of flexion spasms were documented on video. Spasms were characterized by para-axial muscular contraction, diamond-shaped muscle activity on EMG, and ictal high-amplitude slow wave followed by electrodecremental response on EEG.

Dermatologic examination with Wood's light revealed hypomelanotic macules on his fair skin. Ophthalmologic evaluation did not show any lesion. Abdominal and renal ultrasonography were normal. Cardiac exploration revealed three small nonobstructive rhabdomyomas. Brain CT scan and MRI showed calcified subependymal nodules, several cortical tubers, and a large, left frontopolar and basal cortical dysplasia. Epileptic spasms symptomatic of tuberous sclerosis were diagnosed. Genetic analysis documented de novo mutation in the TSC2 gene.

The baby was treated with vigabatrin 100 mg/kg orally per day, which stopped clinical spasms. On the seventh day, background EEG was reorganized, but the left anterior spike focus persisted. Vigabatrin dosage was adjusted to 150 mg/kg per day and permitted to control both clinical manifestations and EEG spikes. The baby promptly recovered his former interaction abilities and made motor progress with head control.

Focal seizures occurred at 4 months of age, characterized by loss of contact sometimes associated with asymmetrical upper limb posture. They initially responded to valproate and vigabatrin combination therapy. At 6 months of age, spasms associated with focal motor seizures reappeared, and the infant exhibited motor and cognitive regression. EEG background was disorganized with multifocal spikes. The seizures remained resistant to antiepileptic drugs. Ketogenic diet was introduced at 12 months of age, with a transient improvement in seizure frequency and acquisition of motor and cognitive milestones. Because of the persistent seizures, he underwent surgical evaluation with subdural grids, followed by a guided left frontal cortical resection. Histology confirmed cortical focal dysplasia and the presence of small hamartomas in white matter. Despite a worthwhile improvement on seizure frequency after surgery, he still experiences at least weekly focal seizures with occasional generalization at 5 years of age. He has moderate to severe mental retardation with autistic features.

Biological basis

Etiology and pathogenesis

Etiology-based classification of infantile spasms distinguishes them as structural, metabolic, infectious, immune, genetic, or unknown etiology (150). It should be noted that infantile spasms from genetic causes that result in structural lesions (ie, tuberous sclerosis complex) are classified under the structural/metabolic category. In contrast, infantile spasms from genetic defects that do not have overt structural/metabolic pathologies may be classified as infantile spasms of genetic cause. Such examples are infantile spasms due to syntaxin binding protein 1 (STXBP1) (16). The term “idiopathic,” previously associated with the fewer cases with normal development before the onset of symmetrical spasms, normal examination, normal imaging, and hypsarrhythmia without focal abnormalities, is no longer used according to the revised ILAE classification (16; 150).

Structural metabolic etiologies include prenatal, perinatal, and postnatal etiologies. The United Kingdom Infantile Spasms Study (UKISS) reported that of 207 infants, 127 (61%) had proven etiology, 68 (33%) had no identified etiology, and 12 (6%) were not fully investigated (129). Etiologies were prenatal in 63, perinatal in 38, postnatal in 8, and other in 18. The most common etiologies were hypoxic-ischemic encephalopathy (21, 10%), chromosomal (16, 8%), malformations (16, 8%), stroke (16, 8%), tuberous sclerosis complex (15, 7%), and periventricular leukomalacia or hemorrhage (11, 5%). The remaining 32 etiologies were all individually uncommon. A study showed that infants with severe hypoxic-ischemic encephalopathy who did not receive therapeutic hypothermia were six times more likely to develop infantile spasms compared to those who did; however, the difference was not statistically significant, though a small sample size may have affected these results (02).

Among the prenatal etiologies, the most common are central nervous system malformations (lissencephaly, hemimegalencephaly, focal cortical dysplasia, septal dysplasia, or callosal agenesis), intrauterine insults, neurocutaneous syndromes (tuberous sclerosis complex, incontinentia pigmenti, Ito syndrome, neurofibromatosis), metabolic disorders (phenylketonuria, tetrahydrobiopterin deficiency, pyridoxine deficiency, and congenital disorders of glycosylation), chromosomal (Down syndrome, del1p36, del2q21.3-q22.2, dup14q12, dup16p11.2, dup15q11q13) or single gene (ARX, STK9, CDKL5 mutations) abnormalities, and mitochondrial pathologies such as PEHO syndrome (progressive encephalopathy, edema of the limbs, hypsarrhythmia and optic atrophy) and early infantile-onset Leigh syndrome (54; 129; 95; 130; 156; 159; 115; 133; 157). In 2013, the Epi4K Consortium and the Epilepsy Phenome/Genome Project (EPGP) reported the data on 264 children with epileptic encephalopathies including infantile spasms. The authors identified new genes implicated in this condition such as GABRB3 and ALG13 (an X-linked encoding a subunit of the uridine diphosphate-N-acetylglucosamine transferase). They also identified mutations affecting genes already reported in epileptic encephalopathy (FLNA, GABRA1, GRIN1, mTOR, NEDD4L). Interestingly, the mTOR mutation (mammalian target of rapamycin) was found in an 18-month-old patient without a brain abnormality, whereas mutations affecting this gene had been previously been associated with hemimegalencephaly (50). Also, Michaud and colleagues studied a cohort of 44 children, and they reported mutations in other genes that had already been associated with other epileptic encephalopathies, including STXBP1, CASK, PNPO, and ADSL (115). Using exome sequencing of 356 trios of patients with epileptic encephalopathies, including infantile spasms, the involvement of de novo mutations of synaptic transmission genes was implicated in the pathogenesis of epileptic encephalopathies (52; 38). These include dynamin 1 (DNM1), which is a GTP-binding protein involved in clathrin-mediated endocytosis. Interestingly, by combining array-CGH and exome sequencing the yield of genetic etiology is much increased: in a study Hino-Fukuyo and colleagues were able to identify novel mutations in 9 of 32 patients, also including two new candidate genes, namely NR2F1 and CACNA2D1 (71). Many more genes continue to be discovered based on individual cases or a small series (KCNQ2, R198Q, FARS2, SPATA5, SLC1A4, GUF1, WDR62, UBA5, SCN2A, PARS2, KCNB1, GNAO1, SLC35A2, TBL1XR1, KIF1A, SEMA5A) (03; 33; 101; 140; 08; 116; 27; 120; 183; 118; 167). The question remains as to why some but not all infants with a mutation experience spasms. The Zika virus, causing congenital Zika syndrome, has been proposed as an etiological agent of infantile spasms (04).

Among the other etiologies, herpetic encephalitis occurring in infants (mean 10 months of age) may represent a risk factor for developing infantile spasms in patients with cerebral lesions involving the insula and the hippocampus (09). Autoimmune etiologies have also been associated with infantile spasms, including folate receptor autoantibodies (155).

Unless the etiology is a specific genetic disorder, such as tuberous sclerosis complex or infantile spasms present in identical twins, familial recurrence is rare. During the past several decades, immunization with various vaccines, especially the diphtheria-pertussis-tetanus vaccine, has been frequently considered as a causative agent in infantile spasms. The relationship is contested because the diphtheria-pertussis-tetanus immunization is given at a time when infantile spasms have their peak occurrence (ie, less than 6 months of age). Some literature reports show evidence indicating that the association between infantile spasms and diphtheria-pertussis-tetanus immunization is coincidental and that the two are not causally related (58; 32; 15; 05). These data can also be supported by data from the work of McIntosh and colleagues, who found no evidence that vaccinations before or after disease onset affect outcome in patients with Dravet syndrome and SCN1A mutations (114).

The pathogenesis of infantile spasms is still unknown. However, it has been hypothesized that there might be a common mechanism by which all of the different etiologies might lead to the same seizure (132). The classical hypothesis suspected that infantile spasms was generated from the brainstem, with projections going both to the cortex to generate hypsarrhythmia and to the spinal cord to generate the spasms (64). Lado and Moshé have proposed that there may be aberrant interactions between the cortex and brainstem (102). Experimental evidence also suggests that spasms may have a focal cortical onset (149; 57; 61; 62).

The following evidence on the role of the cortex has been proposed. Overexpression of axonal collaterals and excitatory synapses that play a major role in the development of cortical functions determine major hyperexcitability of the developing brain cortex and could be responsible for continuous spiking activity, particularly in combination with some brain damage. Lack of myelin at that age may account for the absence of interhemispheric synchrony, thus, producing the hypsarrhythmic pattern (47; 45). Continuous paroxysmal activity would account for the cognitive decline. The ongoing epileptic activity in the cortex should also determine subcortical disinhibition, with paroxysmal discharges in the basal ganglia (31). Thus, a loop including the cortex and basal ganglia would be involved in the genesis of West syndrome (36). Any alteration at one level, either a cortical lesion, which is the most frequent, or basal ganglia dysfunction, particularly in inborn errors of metabolism, may precipitate the dysfunction of this loop. Maturation of the brain would reduce excitability and explain the disappearance of the syndrome in the majority of cases, although under certain, yet unidentified, circumstances may proceed to Lennox-Gastaut syndrome with the occurrence of slow spike waves.

A clinical study has demonstrated temporal lobe impairment by using event related potentials (ERP) in 25 infants affected by infantile spasms. The study demonstrated that ERP latencies decreased with age in controls, but not in patients, and these alterations were not affected by pharmacological treatment (174).

Gale suggested that hypsarrhythmia may represent an ongoing seizure, and infantile spasms and electrodecremental events may result from activation of subcortical circuits attempting to control cortical seizure activity (102). Modifications of the sleep-wake cycle, with reduced REM sleep, are consistent with this hypothesis (79). Brainstem serotonergic neurons are involved in sleep cycles, and depletion of serotonin may decrease REM sleep. Langlais and colleagues provided data supporting a serotonin dysfunction hypothesis by demonstrating reduced levels of 5-hydroxyindoloacetic acid (5-HIAA), a metabolite of serotonin, as well as decreased levels of the catecholamine metabolites homovanillic acid and 3-methoxy-4-hydroxyphenylglycol (MHPG) in patients with infantile spasms, but it is yet undetermined whether this is primary or secondary to infantile spasms (103). In children who responded to ACTH treatment, there was a large increase in 5-HIAA following therapy, whereas in nonresponders, 5-HIAA levels decreased.

Baram has suggested that stress may be implicated based on evidence that the corticotropin-releasing hormone (CRH) is released during stress and may generate convulsions in the developing rat brain (10). In fact, Baram and colleagues demonstrated lower cerebrospinal fluid ACTH, although they failed to demonstrate any difference in cerebrospinal fluid cortisol or CRH levels between infantile spasm patients and controls.

Animal models have been of great help in understanding these mechanisms and in developing new treatments. Current infantile spasm models either focus on a specific cause of infantile spasms or simulate the common behavioral phenotype, ie, spasms. (61). They are divided into acute models of infantile spasms, manifesting spasms only during the acute postinduction period, and chronic models, which recapitulate various aspects of the evolving phenotype of the infantile spasms syndrome.

In the acute models belong the NMDA model of emprosthotonic seizures (111) and the prenatal betamethasone/postnatal NMDA model of infantile spasms (163), as well as the gamma-butyrolactone-induced extensor spasms in the Ts65Dn mouse model of Down syndrome (34). The CRH model has tested the stress theory of infantile spasms and the mechanisms of ACTH effects on seizures, although no spasms have been reported (12). The stress theory of infantile spasms has prompted studies assessing the pathogenic role of CRH on spasms, as a possible mediator of stress-induced infantile spasms. Intraventricular infusion of CRH causes limbic seizures in developing rodents, which are not suppressed by ACTH (12). The authors suggested the alternative possibility that perhaps the therapeutic effects of ACTH are either via suppression of the release of CRH or through activation of melanocortin receptor 4 (MC4R) (23). However, the link between CRH and spasms is still open, as there is no report of spasms induced by intraventricular administration of CRH (12). The NMDA and prenatal betamethasone/postnatal NMDA models exhibit emprosthotonic (forward flexion) seizures acutely following the induction with NMDA (111; 163; 162). A study suggested that although the NMDA-induced emprosthotonic seizures are not ACTH responsive to the tested ACTH doses, prenatal exposure to betamethasone can render them ACTH-responsive (163). However, in a different study, use of higher doses of a porcine ACTH1-39 compound could reduce NMDA-induced spasms (169). Variations of the NMDA model have been reported and include introduction of various methods of prenatal stress (184), postnatal adrenalectomy (169), or methylazoxymethanol-induced malformation (92). In these models, early stressors or adrenalectomy exacerbate spasms induced by NMDA, although there is no chronic epilepsy reported in these models.

Another acute model is the Ts65Dn mouse, a genetic model of Down syndrome, which manifests extensor spasm-like seizures acutely following the injection of gamma-butyrolactone, a prodrug of the GABAB receptor agonist gamma-hydroxybutyric acid (34). This model has been developed based on the evidence that up to 10% of children with Down syndrome develop infantile spasms.

Four chronic models of infantile spasms have been described so far. One is based on the theory of developmental desynchronization (tetrodotoxin (TTX) model) (104), two are mouse models of infantile spasms due to aristaless-related homeobox (ARX) loss of function (112; 138), and the fourth simulates the cortical/subcortical pathology associated with infantile spasms from structural lesions (149).

The TTX model aimed to desynchronize the cortical-hippocampal developmental processes through chronic infusion of TTX, a sodium channel blocker, starting at postnatal day 10 (104). As a result, brief spasms were elicited starting around postnatal day 21, as well as other types of seizures, with interictal EEG abnormalities resembling hypsarrhythmia. In the same model, the analysis of high frequency EEG oscillations (HFOs) suggested that neocortical networks are abnormally excitable, particularly contralateral to the TTX infusion. These abnormalities are not restricted to small areas of cortex. The same study demonstrated that multiunit firing coincided with HFOs, and they both occurred at seizures’ onset, supporting the thesis of a neocortical hyperexcitability hypothesis (57). Kobayashi and colleagues have studied fast oscillations from scalp EEG in patients with hypsarrhythmia. The fast oscillations were very dense before ACTH treatment and were reduced by the treatment. The authors suggested that fast oscillations corresponded to epileptogenicity because of their close relation to the severity of hypsarrhythmia, and they suggested that these high frequencies might affect the process of neurodevelopment (97).

Among the genetic models, the aristaless-related homeobox (ARX) mouse models reproduce the deficit of GABA interneurons due to either conditional deletion of the ARX gene in interneurons (112) or polyalanine triple repeat expansion (ARX plus 7), which results in loss of ARX function (138). These ARX-deficient mice exhibit a variety of seizure types during development, including brief spasm-like seizures resembling infantile spasms, as well as behavioral and neurodevelopmental deficits. Using the ARX plus 7 model, Olivetti and colleagues reported that early postnatal administration of estradiol may prevent epilepsy by restoring the depleted interneuronal populations (126). However, neonatal estradiol had no effect on other models of spasms, such as the multiple-hit rat model of spasms due to structural lesion (63) and the prenatal betamethasone/postnatal NMDA model (25), which may suggest that the estradiol effects may depend upon etiology of infantile spasms.

Scantlebury and colleagues created a chronic model, the multiple-hit rat model of infantile spasms due to structural lesions, by injecting rats with lipopolysaccharide and doxorubicin intracerebrally to obtain cortical and subcortical damage and p-chlorophenylalanine intraperitoneally to deplete serotonin (149). This model reproduces most of the human features, including age specificity, ictal correlates of spasms and interictal epileptiform abnormalities, evolution to other types of seizures, and adverse cognitive consequences. It was shown that in the multiple-hit model there is an underlying parvalbumin-selective interneuronopathy in the cortical regions, which could underlie the epilepsy and dyscognitive phenotype (89). The multiple-hit model has provided evidence for the potential therapeutic benefit of new candidate therapies with potential utility in drug-resistant infantile spasms: high-dose pulse rapamycin effected acute and sustained suppression of spasms and improved cognitive outcome, without significant side effects (139); carisbamate acutely reduced both behavioral spasms and electroclinical spasms during the first 2 to 3 postinjection hours, without detectable toxicity or mortality (128); and a single dose of the galanin analog NAX 5055 had no acute efficacy on spasms but also had no toxicity. This could be due to the low levels of galanin receptor 1 (GalR1) early in life. However, it cannot be excluded that repetitive NAX 5055 administration might show efficacy on spasms (86); CPP-115, a vigabatrin analogue, decreased spasms at considerably lower and better tolerated doses than vigabatrin did in previous studies (22). CPP-115 has shown lower risk than vigabatrin for retinal toxicity in animal studies (131; 152), is currently in phase 1 trial in humans, and has acquired orphan drug indication for infantile spasms by the Federal Drug Administration in the United States. An independent group has reported a case report of an infant with infantile spasms who was transitioned from vigabatrin to CPP-115 treatment with significant improvement in efficacy and tolerability of the drug (40).

Further studies failed to show any efficacy for 17-beta estradiol, the caspase-1 inhibitor VX-765, and the GABAB receptor inhibitor CGP35348 (63).

Pirone and colleagues have proposed a novel adenomatous polyposis colon (APC) conditional knock-out mouse model (137). Their model centers around the ß-catenin pathways in the brain, malfunctions of which have been predicted to be caused by multiple identified infantile spasm risk genes. They showed that conditional deletion in mice of the adenomatous polyposis coli gene (APC cKO), a major regulator of ß-catenin, leads to many important characteristics of human infantile spasms, thus, suggesting new targets for therapeutic interventions.

Despite the high association of infantile spasms with tuberous sclerosis in humans, there have been no reports of spasms in animal models until recently. Acute recordings in a Tsc1(+/-) mouse model revealed electrographic patterns resembling “spasm-like” discharges with high amplitude potentials followed by fast discharges (65). Additional studies are needed to further characterize these events.

Finally, a chronic early stress model of infantile spasms was proposed by Dube and colleagues following chronic intermittent restriction of the bedding and nesting (44).

Epidemiology

The incidence of infantile epileptic spasms syndrome is estimated to be about 2 to 3.5 per 10,000 live births (132). It is one of the most frequent types of epileptic encephalopathy, the group of conditions in which epilepsy determines cognitive deterioration. The onset is usually within the first year of life, with a peak age between 3 and 7 months; onset after 18 months is rare, although onset at up to 4 years of age has been reported (106; 160; 143). Infantile epileptic spasms syndrome occurs in children from all the ethnicities, and boys are slightly more affected than girls (3:2 ratio) (106; 160; 143). Trevathan and colleagues reported a lifetime prevalence at 10 years of age as 1.0 to 2 per 10,000 children (160).

Prevention

There is no known prevention. However, vigabatrin could prevent the occurrence of spasms when given before the onset of spasms in patients with tuberous sclerosis complex (87). After preventative treatment with vigabatrin at 24 months of age, mental retardation was significantly less frequent and less severe compared to children that received treatment at onset of spasms. Also, this prevention scheme led to a higher ratio of seizure-free patients, lower incidence of drug-resistant epilepsy, and lower number of patients requiring polytherapy (88). The EPISTOP trial followed 94 infants with tuberous sclerosis with monthly video EEGs, during which they were given vigabatrin either as conventional antiepileptic treatment after the first electrographic or clinical seizure, or preventively when epileptiform EEG activity was detected before seizures (100). At 24 months, data showed preventive treatment reduced the risk of clinical seizures, drug-resistant epilepsy, and infantile spasms, with no adverse events reported.

Preclinical experimental data suggest that the mTOR inhibitor rapamycin may prevent the occurrence of seizures in a mouse model of tuberous sclerosis complex when given before the onset of seizures, although there is no report of spasms in the existing tuberous sclerosis complex models (186). Evidence in the multiple-hit model of infantile spasms has also supported the efficacy of rapamycin treatment on spasms, given after the onset of spasms in this nongenetic model, and has also demonstrated its potential as disease-modifying treatment (139). However, there is currently no known clinical study testing the efficacy of mTOR inhibitors in human patients with infantile spasms.

Differential diagnosis

Confusing conditions

A number of disorders may mimic infantile spasms and need to be considered in the differential diagnosis. Some of them are benign conditions that usually do not require any treatment; other epileptic conditions may require a different treatment.

Babies with infantile spasms are often misdiagnosed as having exaggerated startle responses. There should be a high degree of suspicion for epileptic spasms if exaggerated startle occurs, especially on arousal.

On occasion, breath-holding spells can be misdiagnosed as infantile spasms. These provoked events are associated with normal interictal EEG. Attacks often spontaneously cease after 5 or 6 years of age and do not require any medical treatment.

Benign myoclonus of early infancy (benign nonepileptic infantile spasms) was first reported in 1977 (105). Although it shares the similar age of onset and behavioral spasms with infantile epileptic spasms syndrome, the prognosis is entirely different. In benign myoclonus of early infancy, the tonic spasms are associated with normal ictal and interictal EEGs during wakefulness and sleep. The spasms occur without any temporal relationship with sleep, in contrast to those of infantile epileptic spasms syndrome. There is no mental or psychomotor involvement. It is usually associated with no or minor perinatal insults. There may or may not be family history of epilepsy (42). The spasms in benign myoclonus of early infancy usually disappear by a few years of age, with or without treatment.

Sandifer syndrome is associated with gastroesophageal reflux, with head cocking, or torticollis, and abnormal dystonic posturing of the body, including opisthotonus. There may be associated eye and limb movements. These spells, particularly the opisthotonic posturing, may be mistaken for spasms. Historical features can help establish the diagnosis, although not all babies with reflux will exhibit obvious signs such as vomiting, failure to thrive, and respiratory symptoms. Spells often occur in relation to feeding. EEG is normal. Barium esophagogram, esophagoscopy, or pH probe may demonstrate the reflux. The major risk is to overlook spasms in a child with reflux, a frequent condition in early infancy. Any doubt should indicate an EEG.

Among the epileptic conditions, Ohtahara syndrome and benign myoclonic epilepsy need to be taken into account. Ohtahara syndrome is a severe epileptic encephalopathy with frequent tonic spasms, but in comparison to infantile spasms, it has an earlier onset in the neonatal period and exhibits burst-suppression on interictal EEG (125). Benign myoclonic epilepsy of infancy is another epilepsy syndrome with a similar age of onset, but a distinct seizure type. The latter are characterized by myoclonic jerks, usually involving only the arms and head. As in benign nonepileptic infantile spasms, the myoclonic episodes usually have no relationship to sleep, although drowsiness tends to increase their frequency. Interictal EEGs are usually normal, but the seizures are associated with a 1- to 3-second burst of spike-and-wave and polyspike-and-wave discharges. Photic stimulation may provoke the myoclonus. Psychomotor development remains normal, and these children do not evolve to have other seizure types.

Diagnostic workup

The evaluation begins with the history and physical examination of the patient in order to collect the entire description of the episodes and information about the neurologic status of the little patient. Sometimes a home video can be of great help. The diagnostic workup continues with the EEG evaluation, aimed to assess the presence of ictal phenomenon and interictal hypsarrhythmia. It is recommended that an EEG evaluation is conducted as soon as possible after the onset of clinical spasms. This should be a full EEG evaluation, including a complete sleep-wake cycle assessment. An overnight inpatient 24-hour video-EEG is the best choice; otherwise, a prolonged 2- to 4-hour EEG during a waking and sleep period may be sufficient (109). Hypsarrhythmic pattern is most frequent during non-REM sleep followed by waking and arousal, and spasms are better captured on arousal (76). A thorough search for etiology should be conducted. In children younger than 2 years old, MRI including 3-dimensional (3D) T1-weighted gradient-recalled echo sequence, axial and coronal T2, and fluid-attenuated inversion recovery (FLAIR) sequences is recommended. In children younger than a year old, FLAIR and 3D T1-weighted gradient-recalled echo sequence is less helpful (60). Magnetic resonance spectroscopy (MRS) may be helpful in the hypothesis of inborn errors of metabolisms. Repeated imaging is recommended in children not responding to treatment or in cases with clinical deterioration. In fact, focal cortical dysplasia may not be detectable until maturation of myelinization has been completed (by the age of 24 to 30 months) (60). In some children, fluorodeoxyglucose-positron emission tomography (FDG-PET) is recommended when there is evidence for focality, but MRI imaging is negative (60). FDG-PET studies have revealed focal areas of hypometabolism, which often correlate with dysplastic cortex and white matter (29; 30). Newer modalities including magnetoencephalography (MEG), subdural grid electrodes, and stereoelectroencephalography (SEEG) may be used in suspected focal cases with difficult to localize lesions (01).

If a cause has not yet been identified, workup should proceed by looking for a metabolic origin. Pyridoxine dosed at 100 mg iv may be administered to screen for pyridoxine-dependent seizures. Also, urine for organic acids, biotinidase determination, neurotransmitter CSF evaluation, lactic acid, and folate should be considered. In parallel, testing for gene mutations or rare chromosomal disorders, such tuberous sclerosis complex mutations or ARX in males and CDKL5 in females, may also be considered in patients with suggestive characteristics (132).

In light of new genetic findings, the PERC (Pediatric Epilepsy Research) group proposed that a cost-effective workup for those without obvious cause after initial clinical evaluation and MRI includes an array-CGH followed by an epilepsy gene panel if the microarray is not definitive, serum lactate, and urine organic acids (51; 178).

Management

Drug treatment. According to the U.S. consensus report of the Infantile Spasms Working Group (ISWG), effective treatment for infantile spasms should produce both cessation of spasms and resolution of hypsarrhythmia on EEG and is an ‘‘all-or-none’’ response (109; 110; 132). This effect has also been associated with the best cognitive and developmental outcomes, including reduced progression to other seizures disorders. Effective short-term treatment should be pursued when possible to avoid side effects. Timely management of patients who are refractory to first-line treatment is critical in order to avoid long-term consequences (76; 109; 144). In addition, many patients with infantile spasms may develop other forms of epilepsy over time, requiring conventional antiepileptic drugs (110).

The antiepileptic drugs most used to cure spasms are ACTH, oral steroids, and vigabatrin. Other antiepileptic drugs used are valproic acid and nitrazepam. Data are lacking on the best approach to take if spasms recur following an initial clinical response to treatment (ie, relapse). Possible treatment options include returning to the previously effective treatment agent and dose protocol, returning to the previously effective treatment agent but at the maximum dose, or implementing a new treatment agent (132).

Treatment with the maximum dose of ACTH (150 IU/m² per day bid) for two weeks, followed by careful taper and evaluation of treatment response is used (11), although there are no comparisons in this study with other doses such as the commonly used dose of 100 IU/m² daily for 4 weeks (179). The treatment effect of ACTH has a rapid onset, with a mean time to treatment response of two days (11). It has been shown that the effectiveness of once-daily dosing of high-dose ACTH may be comparable to twice-daily dosing (72). The all-or-none resolution of both spasms and hypsarrhythmia has suggested a possible disease-modifying effect of high-dose ACTH in a few patients (154). However, despite the beneficial effects of current therapies, including ACTH, the concerns over the high relapse rates, the occurrence of other types of seizures, and the still significant residual neurodevelopmental deficits in patients with infantile spasms underlie the importance of finding better disease-modifying therapies.

Hrachovy and colleagues compared high-dose and low-dose natural ACTH (77). The high dose was 150 IU/m² per day for 3 weeks followed by 80 IU/m² per day for 2 weeks, 80 IU/m² every other day for 3 weeks, 50 IU/m² per day every other day for 1 week, with dosage then tapered to zero during a 3-week period. The low-dose group received 20 IU/day for 2 weeks, which was then increased to 30 IU/day for 4 weeks if there was no response. There were no differences in efficacy for each dosing group, with 50% response in the high-dose group and 58% response in the low-dose group. The adverse effect profile (irritability, increased appetite) was similar except for a higher rate of hypertension in the high-dose group. Another study showed that the efficacy of low dose synthetic ACTH therapy without tapering for treatment of infantile spasms was comparable to that of other reports using synthetic ACTH therapy followed by a taper (98).

Mytinger and colleagues reported their experience with intravenous methylprednisolone for the treatment of infantile spasms. A pulse dose of 20 mg/kg intravenous methylprednisolone on each of three successive days, followed by a 2-month oral prednisolone taper, led to the rapid remission of infantile spasms in 50% of the treated infants. The authors also estimated the cost to be cheaper than a typical course of ACTH, suggesting that intravenous methylprednisolone and oral corticosteroids is a reasonable cost-effective approach to infantile spasms (119). A subsequent study failed to demonstrate superior rates of electroencephalographic or clinical remission in patients treated with synthetic adrenocorticotropic hormone (40 to 60 IU/every other day) compared with prednisolone (40 to 60 mg/day). In the same study, more patients achieved electroclinical remission when treated with prednisolone than with adrenocorticotropic hormone (171). In a long-term follow-up study, the investigators found that control of spasms at three months was significantly better in the prednisolone group than the ACTH group; however, at 6 and 12 months control of spasms was not significantly different despite a trend favoring prednisolone. The risk of relapse following initial remission was similar in the two groups (170).

Vigabatrin (gamma vinyl GABA), an antiepileptic medication available in Canada, Europe, South America, several countries in Asia, and the Middle East, has been used with success in the treatment of infantile spasms, as confirmed in two double-blind studies (07; 49). Two controlled studies comparing vigabatrin to steroids [vigabatrin 100 to 150 mg/kg per day; synthetic ACTH1-24 40 to 60 IU (0.5 to 0.75 mg) alternate days] in patients with all other conditions than tuberous sclerosis found better short-term effect of steroids than vigabatrin (165; 107). However, the rate of relapses is higher with steroids, and tolerability is better with vigabatrin (165; 107). In addition, the response at 12 to 14 months in the two groups (hormonal treatment vs. vigabatrin) showed no difference (108). Similarly, Mohamed and colleagues found a 61.1% response rate for steroid therapy and a 42.5% response rate for vigabatrin in a retrospective case study over eight years of children with infantile spasms (117). Both groups had a similar relapse rate. A randomized study has shown better effect of vigabatrin than hydrocortisone in infantile spasms due to tuberous sclerosis (26). The constriction of visual field reported with vigabatrin raised concerns about its use (48). Other studies, however, show a low risk of developing this side effect at short term (6 months); therefore, vigabatrin is an appropriate option for patients with infantile spasms who receive a clinical benefit from its effectiveness, given the clinical consequences of uncontrolled spasms (177; 176). Besides, some emerging experimental data suggest new strategies in order to minimize this side effect (84). These data allowed the mid-2009 FDA approval for the adjunctive treatment of refractory complex partial seizures and as treatment of infantile spasms. The duration of vigabatrin use should be limited due to its potential side effects; however, the optimal length of treatment is unknown. One study found that a shortened vigabatrin course of 6 months could be sufficient to treat and prevent relapse of spasms in children with infantile spasms, particularly those without an identified etiology (37). Doumlele and colleagues presented a child with infantile spasms treated with CPP-115, a high-affinity vigabatrin analogue, through an investigational new drug protocol (40). He experienced a marked reduction of seizures with no evidence of retinal dysfunction. Thus, CPP-115 may be a promising alternative to vigabatrin.

A randomized, open-label multicenter trial sought to investigate the efficacy of hormonal treatment (prednisolone or tetracosactide depot) versus hormonal treatment with vigabatrin for infantile spasms (122). Prednisolone was given orally (10 mg, 4 times a day) or tetracosactide depot was given intramuscularly (0.5 mg [40 IU] on alternate days) for 2 weeks. If this was ineffective, the dose was increased (prednisolone 20 mg, 3 times a day or tetracosactide depot 0.75 mg on alternate days). Hormonal therapy was tapered after 2 weeks. Vigabatrin was given orally in 2 doses per day, starting at 50 mg/kg per day and increasing to 100 mg/kg per day. If spasms continued, this was further increased to 150 mg/kg per day. The primary outcome was cessation of spasms, defined as “no witnessed spasms as recorded by parents and carers on and between day 14 and 42.” Results showed that the primary outcome was reached in 72% of patients on hormonal therapy with vigabatrin, versus 57% of patients on hormonal therapy alone, thus, showing that combination therapy is significantly more effective than hormonal therapy alone. An 18-month follow-up study showed that combination therapy did not result in improved developmental or epilepsy outcomes (123). Furthermore, acute encephalopathy with extrapyramidal symptoms, vigabatrin-associated brain abnormalities on magnetic resonance imaging, and death in one patient has been reported shortly after initiation of therapy with vigabatrin and ACTH (17).

The Guideline Development Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society reported the following recommendations based on a Pubmed and Embase research from 2001 to 2011 (67):

Previously, among antiepileptic drugs only valproic acid and nitrazepam were reported to be effective in treating patients with infantile spasms. Dreifuss and colleagues have suggested that nitrazepam and ACTH afford a similar degree of seizure control, although there is no general agreement (43). A study suggested that infantile spasms respond poorly to topiramate (173). This was supported by a more recent study that found adding on moderate dose topiramate to high dose prednisone therapy provided no benefit in achieving spasm freedom (182). Zonisamide has also been shown to have limited efficacy among patients who do not respond to first-line therapy (81). Olson and colleagues treated 107 children suffering from infantile spasms with rufinamide as an adjunctive therapy to a median number of three antiepileptic drugs (127). Responder rate was 53%, and the median reduction in spasm frequency was 50%. Overall, rufinamide was well tolerated. Further prospective studies are warranted to validate this observation (154). Clobazam was studied as adjunctive treatment and it was found that about 22% of patients became spasm free for at least two weeks after initiation of clobazam, 34% of which remained spasm free until the last follow-up and did not require the administration of other antiepileptic drugs (69). Felbamate controlled epileptic spasms in 8 of 29 infants resistant to first-line treatment in a small retrospective study (41). Cannabidiol was not found to be particularly effective in highly refractory cases, but may be effective in younger patients with shorter duration of spasms, based on a small study of nine patients (80). Everolimus was approved for the adjunctive treatment of refractory tuberous sclerosis complex-associated focal seizures, and a small single-center prospective observational study showed it appears to have the potential to treat both spasms and hypsarrhythmia in infants with tuberous sclerosis complex-associated West syndrome (147).

Data report a possible response of infantile spasms to ketogenic diet (99). However, it is suggested to consider it only for infantile spasms refractory to first line treatment, after failure of corticosteroids and vigabatrin (74; 136). In another study, Hussain and colleagues reported a limited efficacy of the ketogenic diet in the treatment of highly refractory infantile spasms (only 2 of 22 patients achieved a complete response and the success was more reasonably attributable to alternative therapies in both cases) (82). The modified Atkins diet was found to be effective and well tolerated in children with infantile spasms refractory to hormonal therapy (151).

The National Infantile Spasms Consortium has evaluated 330 infants with newly diagnosed infantile spasms in a multicenter study. They considered ACTH, oral corticosteroids, and vigabatrin as standard treatments and analyzed them individually, whereas all other nonstandard therapies were analyzed collectively. Overall, 46% of children receiving standard therapy responded compared to only 9% who responded to nonstandard therapy (p < 0.001): 55% of infants receiving ACTH as initial treatment responded, compared to 39% for oral corticosteroids, 36% for vigabatrin, and 9% for other. Neither etiology nor development significantly modified the response pattern by treatment group (96). In another study performed by Knupp and colleagues, using data from the same consortium, investigating response to second treatment after initial failed treatment, which included 118 infants, overall response rate to a second treatment was 37%, with children who received standard medications (compared to all other nonstandard treatments) with differing mechanisms for first and second treatment having a higher response rate than other sequences (55% vs. 25%) (96). The authors also found that those children who received the first treatment within four weeks of infantile spasm onset had a higher response rate to a second treatment as compared to those initially treated after this time period.

Other strategies have been used in the past; however, they have been reported anecdotally. In uncontrolled studies, human polyvalent immunoglobulins have been used intravenously to decrease the frequency of seizures, improve EEG patterns, as well as improve psychomotor performance (161).

High-dose pyridoxine (100 to 300 mg/kg per day) has been beneficial in treating some patients with infantile spasms, with minimal toxicity (19; 134). Those patients who respond tend to do so within the first 1 to 2 weeks after initiation. In preclinical studies, rapamycin has been evaluated for tuberous sclerosis complex in animal models (186) or the multiple-hit model of non-tuberous sclerosis symptomatic infantile spasms (139), and it can be considered an emerging therapy under investigation for infantile spasms. In the same multiple-hit model of infantile spasms, carisbamate has also been proven acutely effective in suppressing spasms (128).

Ganaxolone, a neurosteroid that allosterically modulates and enhances tonic GABAA receptor activity has also been tested in an open-label, add-on trial of 20 children 7 months to 7 years of age with drug-resistant infantile spasms (91). Ganaxolone reduced spasms by at least 50% in a third of the children and improved spasms (25%-50%) in another third. One child became spasm-free but did eventually develop astatic seizures. Ganaxolone was also tested in the prenatal betamethasone/postnatal NMDA model of infantile spasms (163), where it significantly delayed the onset and reduced the number of spasms compared with controls (185). In the ARX plus 7 mouse model of spasms, early neonatal administration of 17β-estradiol improved the interneuronopathy and prevented epilepsy (126). However, neonatal 17β-estradiol had no effect in the multiple-hit model of infantile spasms (63) and the prenatal betamethasone/postnatal NMDA model (25).

Surgery. Some infants with medically intractable infantile spasms and focal lesions may benefit from resection of the focal abnormality (31; 73). Curative epilepsy surgery is best accomplished at an early age and in those patients with concordant lesional abnormalities on MRI and EEG (28). Identification of the epileptogenic zone in some of these patients can be challenging, especially when there is no identifiable lesion on MRI. Advances in neuroimaging and invasive monitoring have led to the ability to surgically treat many of these patients who were previously not considered surgical candidates (01). Persistent spasms not amenable to focal surgery and that suffer from drop attacks may benefit from total callosotomy, whereas anterior callosotomy is ineffective, probably for reasons related to maturation of the brain (135).

Outcomes

Outcome may depend on many factors. However, etiology appears to be the most important determinant: a known etiology is a predictor of poor developmental outcome, whereas bilateral/diffuse brain lesions predict both poor development and seizures (164). Cases with infantile spasms associated with negative results on imaging may have a more favorable outcome (132). Among 22 patients with infantile spasms with negative imaging and early effective treatment (within 1 month of infantile spasm onset), normal cognitive development was documented in all 22 patients during a 6- to 21-year follow-up period (94). A review of 67 published studies with an average follow-up period of 31 months found that only 16% of patients with infantile spasms had normal development (76). Poor outcome included mortality, presence of seizures disorders, and cognitive and developmental problems. Classification into the unknown etiology category was again associated with good prognostic outcomes. Other prognostic indicators, including features of normal prior development, absence of causative features, normal imaging studies, absence of other seizures types, and sustained response to therapy without relapse were also associated with favorable prognostic outcomes (76; 145). Initial presentations of epileptic spasms with hypotonia or developmental delay most strongly predict both seizures and neurodevelopmental outcomes (164). Autism spectrum disorder is usually observed in children with symptomatic infantile spasms. Early diagnosis and treatment do not prevent autism spectrum disorder as an outcome of infantile spasms (18).

The National Infantile Spasms Consortium compared outcomes and phenotypic features of patients with infantile spasms with and without hypsarrhythmia and found that first line treatment with standard therapy (which included ACTH, prednisolone, or vigabatrin) was the most important variable in determining likelihood of response to treatment in patients with or without hypsarrhythmia (35). Although response to treatment is associated with a better outcome, it is unclear whether response to one specific treatment is favorable over another. Riikonen found that a good response to ACTH is associated with good neurologic outcome (141). A similar result has been reported for vigabatrin (39), and these investigators state that the long-term outcome in these patients was comparable to that in patients treated with ACTH or steroids, as reported in earlier studies. Automatic detection of interictal ripples on scalp EEG have been used to evaluate the effect and prognosis of ACTH in patient with infantile spasms. One study found that after ACTH treatment, the number and channels of ripples were significantly lower, whereas the percentage decrease in the number, spectral power, and channels of ripples was significantly higher in the seizure-free group than in the nonseizure-free group (168).

The spasms and hypsarrhythmic EEG tend to disappear spontaneously before three years of age. However, up to 55% to 60% of children with infantile spasms will develop other types of seizures and epileptic syndromes (85; 113; 141). Reports from the literature show that 95% of infantile spasm patients have active epilepsy at 10 years of age; 50% of children with infantile spasms develop Lennox-Gastaut syndrome before 11 years of age, and a history of infantile spasms was found in approximately 39% of children with Lennox-Gastaut syndrome (160).

Infantile spasms constitute the most frequent of all seizure types in infants with Down syndrome. Sanmaneechai and colleagues studied 12 infants with Down syndrome, seven of which demonstrated a complete response to high-dose natural adrenocorticotrophic hormone, but four had relapses, which occurred as long as two years after the first event. This suggests that close follow up is necessary even after successful initial treatment (148). Another study found that those with Down syndrome and infantile spasms had a lower risk of subsequent epilepsy following infantile spasms after a 22-month median follow-up time than those with an unknown etiology (13). Neurodevelopmental outcome in patients with spasms associated with Down syndrome is poor; however, the delay in treatment does not appear to contribute to any differences in their developmental scores (158).

Special considerations

Anesthesia

Precautions must be taken regarding seizures.