Demographic characteristics and clinical presentation of infants with infantile epileptic spasms syndrome and their response to therapy: Data from Sri Lanka Infantile Spasms Registry

Abstract

Objectives

Infantile epileptic spasm syndrome (IESS) is an epileptic encephalopathy with often devastating developmental consequences. Most children with IESS have a known etiology, although differing in proportion by geographical settings. Therefore, registries are useful to understand the characteristics of IESS in different countries. The Sri Lanka Infantile Spasms Registry (SLISR) was established to study the demographics and etiology of infants with IESS and their response to therapy in a resource‐limited country.

Methods

Five pediatric neurologists (out of nine) in different parts of the country prospectively recruited children with IESS. The etiology was evaluated using the services available in each setting. Response to treatment for standard (adrenal corticotropic hormone, prednisolone, or vigabatrin) versus nonstandard medications was evaluated at two and six weeks.

Results

Included in the current analysis were 270 children who were registered since 2017. Median age at presentation was 5.36 months (SD 3.6). The mean interval between seizure onset and treatment onset was 1.7 months (SD 1.3). A sizable proportion of the children (61.2%) did not complete the evaluation of etiology. Structural brain abnormality was the most frequently identified etiology in those who were evaluated (38.8%); hypoxic‐ischemic injury was the most common antecedent. The majority of the patients (86%) received a recommended standard therapy as the first treatment, with prednisolone being the most frequent choice. By treatment day 14, the first treatment had achieved spasm control in 63.8% and an electro‐clinical response in 43.6%. While both standard therapies led to positive outcomes, oral prednisolone produced the best therapeutic response.

Conclusion

We describe the etiologies, treatment choices, and response to first‐line medications in a large group of children with IESS from a South Asian country. Although most patients received a recommended first‐line therapy (most often prednisolone), a sizable number initially received nonstandard therapy. Our data illustrate the challenges in the management of IESS in a resource‐limited environment.

Introduction

Infantile epileptic spasms syndrome (IESS), previously known as West syndrome, was first described by William West in an 1841 letter to The Lancet. ¹ This description included most of the clinical features known to us including its refractoriness to therapy and long‐term poor outcome. It is now described as the most common developmental and epileptic encephalopathy, presenting in infancy, with an average prevalence of 30 per 100 000. ² This estimate may not reflect the true disease burden worldwide, since population‐based data on IESS are reported mainly from high‐income countries. Patient registries are useful to understand the important clinical and epidemiological features of IESS because of its rarity. The National Consortium for Infantile Spasms in the United States, a large‐scale multicenter registry of patients with IESS, is an example that generated useful data on IESS etiology and response to therapy. ³

The etiology of IESS is broadly classified into two groups: known and unknown. The known etiological subtypes are roughly similar to those outlined in the International League Against Epilepsy (ILAE) 2017 classification of epilepsies. ⁴ Considering the wide range of etiologies of IESS, regional differences in their prevalence are very likely, for example, the differences in the proportions of etiological subtypes reported from the United States ³ and those reported from India. ⁵ Understanding these regional differences ⁶ is crucial since etiology is the single most important factor associated with better prognosis. ⁷ The absence of a known etiology is associated with better cognitive outcomes as shown in several studies. ⁸ Hence, the differences in reports on the response to therapy in different settings could be related to higher proportions of children with unknown etiology in higher‐income countries ³ versus predominant known etiologies in lower‐income countries. ⁹ Apart from etiology, other factors such as early treatment of spasms is related to better development in later life. ¹⁰ Early onset of spasms has been described as a poor prognostic factor by the same group of investigators, but the small number of patients (n = 10) in this study limits the generalizability of its findings. ¹⁰ Since successful spasm control predicts better long‐term outcomes, it is important to investigate the factors that may affect therapeutic outcomes, such as age of onset of spasms and delay in commencing therapy.

We describe the clinical features, etiology, first‐line treatment, and response to therapy (standard versus nonstandard) of a large group of children with IESS in a resource‐limited setting. We also analyze the association between age of onset and delay in treatment on early control of spasms.

Method

This is a prospective observational study of children with epileptic spasms diagnosed and followed up by pediatric neurologists in Sri Lanka. The Sri Lanka Infantile Spasms Registry (SLISR) was established in 2016 to create a database of children with epileptic spasms in the country. Five of the nine pediatric neurologists serving in different tertiary care hospitals in the country took part by registering children with a confirmed diagnosis of infantile epileptic spasm syndrome.

Patient characteristics, details of clinical presentation including age at onset of spasms and age at diagnosis, initial electroencephalography (EEG) findings, the initial treatment offered, and the response to therapy were fed into a common database. The clinical presentation included details about the spasms (single or clusters), concurrent other seizure types, delay in diagnosis (measured in weeks), development (at the time of onset of spasms), and developmental regression. The EEG findings were noted as the presence or absence of hypsarrhythmia and/or other abnormalities based on a 30‐min sleep EEG followed by five min of wakefulness, interpreted by a child neurologist. Onset age was documented in months to the nearest week. Onset age was categorized as early if the onset was less than 3 months and late if the onset occurred beyond 24 months. Treatment lag was documented as short (<14 days) and long (>28 days). The first‐line treatment offered was categorized as standard first‐line therapy if treated with either hormonal therapy (intramuscular long‐acting adrenocorticotropic hormone [ACTH] or oral prednisolone) or vigabatrin. Any other therapy (antiseizure medication or ketogenic diet) given as first‐line therapy was considered nonstandard. Response to therapy was assessed for spasm cessation and electrical remission. If there were no spasms for more than 48 h by day 14 and the EEG did not show a hypsarrhythmia when reviewed between 14 and 28 days, this was considered an electroclinical response. In patients who could not attend posttreatment EEG recording, the response to therapy was reported as a clinical response only. In those who did not respond by 14 days but whose spasms resolved later, the time taken for a response was documented as between 14 and 28 days or as >28 days.

While the choice of the first medication for spasms was usually determined independently by the attending pediatric neurologist, in some cases therapy was determined by the referring pediatrician. The doses for hormonal therapies were similar in every center (40–60 mg/day for oral prednisolone and 40–60 IU every other day for synthetic adrenocorticotrophic hormone [ACTH] given intramuscularly) as per the treatment schedule used in the Sri Lanka Infantile Spasm Study. ¹¹ The duration for hormonal therapy was 14 days followed by a slow taper over three weeks. The use of vigabatrin was limited mainly to individuals with tuberous sclerosis complex, at a dose of 100–150 mg/kg/day. ¹² The duration of therapy was for a minimum of three months. The use of nonstandard therapy was at the discretion of the attending pediatric neurologist or pediatrician who referred the patient. The doses for these nonstandard therapies mirrored the starting doses proposed in the pediatric British National Formulary. The duration of therapy with these medications was not standardized.

This study was approved by the Ethics Review Committee of the Faculty of Medicine, Colombo, Sri Lanka.

Results

There were 270 patients registered in the SLISR (Mar 2022) by five pediatric neurologists out of nine in the country. Most patients (204, or 75.6%) were registered from a center based in the capital, where two pediatric neurologists recruited patients (64.5% [JW] and 11.1% [PR]). The number of patients registered from other centers in descending frequencies was 30 (11.1%), eight (3%), and seven (2.6%). The demographic features are shown in Table 1. There was a male predominance of 1:1.45. The mean age of spasm onset was 5.43 months (SD 3.63). The mean age at diagnosis was 6.84 (SD 3.62) months, and the mean treatment lag was 1.59 (SD 2.19) months. The age at onset was early (<3 months) in 77 children (28.5%). There were no children with onset later than 24 months. The lead time to commencement of therapy was short (<14 days) in 95 (35.2%) but delayed (>28 days) in the majority (125; 46.3%). The majority (74%) experienced spasms as clusters, but 14% had experienced other types of seizures beyond the neonatal period, before the onset of spasms. Two‐thirds of the patients experienced delayed development even before the onset of spasms. All babies had nonrapid eye movement sleep captured during the EEGs and 90% of the cohort demonstrated hypsarrhythmia in these EEGs.

Table 1.

Demographic features of children diagnosed and treated for infantile epileptic spasm syndrome.

Baseline characteristics	Findings (n = 270) N (%)
Demographic features
Male	160 (59.3)
Ethnicity
Sinhalese	229 (84.8)
Muslim	24 (8.9)
Tamil	17 (6.3)
Mean age at onset in months ± SD (IQR)	5.43 ± 3.63 (3.0–7.0)
Mean age at initiation of treatment in months ± SD (IQR)	6.84 ± 3.62 (4.5–8.5)
Delay in diagnosis in months ± (IQR)	1.59 ± 2.19 (0.25–1.5)
Clinical details
Type of spasms
Single	6 (25.7)
Clusters	131 (49.4)
Both	66 (24.9)
Concurrent seizures with the onset of spasms	26 (9.6)
Seizures prior to onset of spasms	39 (14.4)
Developmental delay at onset of spasms	174 (64.4)
Developmental regression due to spasms
Present	80 (29.6)
Absent	85 (31.5)
Not clearly assessed	104 (38.5)
EEG diagnosis
Hypsarrhythmia	242 (89.6)
No hypsarrhythmia but other abnormalities present	20 (7.4)
Not done/not available	8 (3.0)
Normal	0 (0)
Etiology
Structural	105 (38.8)
Genetic	9 (3.3)
Metabolic	3 (1.11)
Incomplete evaluation	136 (50.4)
Unknown	17 (6.3)
Treatment details
Standard first‐line therapy	231 (85.6)
Nonstandard therapies	39 (14.4)

Abbreviations: EEG, electroencephalography; IQR, interquartile range.

A likely etiology was confirmed in only 117 (43.3%) children. Structural etiologies dominated (38.8%). Hypoxic‐ischemic encephalopathy was the single most frequently identified etiology (38.7%), followed by injury related to central nervous system infections (16.2%). The majority of the children (136/270; 50.3%) had no confirmed etiology due to difficulties in access to or the nonavailability of or incomplete follow‐up on recommended minimal investigations. Restricted access to magnetic resonance imaging resulted in computerized tomography and ultrasound scanning being performed as the imaging modality in 79 and 24 patients, respectively. Nonavailability of screening for metabolic and genetic diseases in the state health system accounted for the limited confirmation of genetic or metabolic etiologies. Only two children underwent whole exome sequencing. Therefore, only 6.3% of children were confirmed to have no known etiology after complete investigations following the recommended benchmark for investigations for etiology. ¹⁰ The identified etiologies are shown in Table 2.

Table 2.

Identified etiologies responsible for infantile epileptic spasm syndrome (n = 117).

Etiology	Number (%)
Genetic
Tuberous sclerosis complex	4 (3.44)
Aicardi syndrome	1 (0.86)
Trisomy 21	2 (1.72)
SCN2A mutation	1 (0.86)
Neurofibromatosis	1 (0.86)
Metabolic
GLUT1 deficiency	1 (0.86)
Biotinidase deficiency	2 (1.72)
Structural
Hypoxic ischemic encephalopathy	45 (38.7)
Central nervous system infection‐related injury	19 (16.3)
Congenital infection	8 (6.89)
Perinatal arterial ischemic stroke	8 (6.89)
Prematurity‐related white matter injury	5 (4.31)
Intracranial hemorrhage	5 (4.31)
Hypoglycemia‐related injury	6 (5.17)
Corpus callosum agenesis/dysgenesis	4 (3.44)
Pachygyria	2 (1.72)
Nonspecific cerebral atrophy	2 (0.86)
Gross hydrocephalus	2 (0.86)

Treatment

Two hundred and thirty‐one (86%) patients commenced treatment with one of the recommended standard first‐line therapies. This included oral prednisolone being given to 171 (63.3%) patients, intramuscular synthetic ACTH to 54 (20%) patients, and vigabatrin to six (2.2%) patients. The remaining 38 (14%) were initially begun on nonstandard oral antiseizure therapy. The antiseizure medications given included sodium valproate in 23 (8.5%), clonazepam in seven (2.6%), carbamazepine in four (1.5%), phenobarbitone in two (0.7%), and levetiracetam in two (0.7%). None were given pyridoxine or a ketogenic diet as the standard first‐line therapy. The 39 children receiving nonstandard therapy included five (12.8%) who had no hypsarrhythmia, 14 (35.9%) with very low body weight, nine (23.1%) who were less than 3 months of age, one with concomitant myoclonic seizure, and trisomy 21 in one child. A comparison of clinical variables between the two groups commenced on standard and nonstandard first‐line medications is shown in Table 3. There was no difference in the clinical variables between the two groups except for a larger group of those on the standard therapy not having had a complete evaluation of their etiologies.

Table 3.

Comparison of clinical characteristics, lead time to treatment, etiology, EEG finding, and development in those treated with standard versus nonstandard therapies.

Clinical variable	Standard treatment (n = 231)	Nonstandard treatment (n = 39)	p value
Male sex	138	22	0.695
Mean age at onset in months ± SD (IQR)	5.38 ± 3.58 (3.0, 7.0)	5.69 ± 4.00 (2.5, 7.13)	0.642
Mean age at initiation of treatment in months ± SD (IQR)	6.84 ± 3.49 (4.5, 8.5)	6.83 ± 4.40 (3.44, 8.25)	0.98
Lead time to treatment (valid %)			0.775
<14 days	80 (29.6)	15 (5.6)
14–28 days	42 (15.6)	8 (3)
>28 days	109 (40.4)	16 (5.9)
Etiology (n, %)			0.007
Structural	85 (31.5)	20 (7.4)
Metabolic	2 (0.7)	1 (0.37)
Genetic	5 (1.9)	4 (1.5)
Unknown	14 (5.2)	3 (1.1)
Incomplete evaluation	125 (46.3)	11 (4.1)
EEG at presentation (n, %)			0.245
Hypsarrhythmia	210 (77.8)	32 (11.9)
Normal EEG/other abnormalities	15 (5.6)	5 (1.9)
EEG not done/unavailable	6 (2.2)	2 (0.7)
Development			0.5
Delayed	147 (54.4	27 (10)
Not delayed	84 (31.1)	11 (4.4)
Concurrent seizures with the onset of spasms			0.188
Present	20 (7.4)	6 (2.2)
Absent	211 (78.1)	33 (12.2)

Abbreviations: EEG, electroencephalography; IQR, interquartile range.

Response to therapies

The subsequent follow‐up to describe the response to therapy was available in only 227 (84%) patients. The overall rate of spasm cessation within 14 days of treatment was seen in 61.2% (n = 227). A further 18.5% achieved control between the 14th and 42nd days with either type of therapy, increasing the cumulative response rate by six weeks to 79.7%. If they were treated with standard first‐line therapy (187), the day 14 response rate increased to 70.5% and overall response by six weeks to 88.7%. The posttreatment EEG on the 14th day was performed or was available for review in only 176 (77.5%) children. Forty‐five percent of them had resolution of the previously recorded hypsarrhythmia with either standard or nonstandard therapies. Thus, only 33% of the total group achieved electroclinical response by day 14. In those children lost for EEG follow‐up, the majority (84%) were in those given standard therapy (n = 43).

When cessation of spasms by the 14th day (clinical response) was the assessed outcome measure, initial treatment with standard therapy achieved spasm control in 132 (63.8%) compared with 7 (35%) in those given nonstandard therapy (p = 0.004). In the 176 children who had posttreatment EEGs, the proportion achieving electroclinical response was 46.3% with standard therapy versus 0% in those given nonstandard therapy (p = 0.001). The total number who achieved spasm control on day 14 was 132. Prednisolone accounted for 80.3% of the individuals with spasm control, while ACTH and vigabatrin accounted for 15.9% and 3.8%, respectively. Seizure freedom was achieved in 100% of the children taking vigabatrin (n = 5), 75.6% for those on prednisolone, and 51.5% for those who received ACTH. In the non‐tuberous sclerosis patients, spasm control on day 14 was significantly higher in those treated with prednisolone (p = 0.003). The early onset of epileptic spasms (<3 months of age) did not influence the control of spasms on day 14 with standard therapy (p = 0.69). Similarly, there was no significant difference in the spasm control by the 14th day between those who commenced therapy early (<14 days from onset) versus late (>14 days) (p = 0.35) or had normal versus delayed development at the onset of spasms (p = 0.49). There was no significant difference in the response rate between those who had a clearly identified etiology versus those incompletely evaluated or those with no clear etiology (p = 0.07).

Discussion

Using a large IESS patient registry from a resource‐limited South Asian country, we describe a range of variables related to the demographics, clinical presentation, and treatment response of infants with IESS and these infants' responses to treatment. Our data illustrate the paucity of information about etiology of IESS in resource‐limited settings in low‐ and middle‐income countries. Despite these resource limitations, a high proportion of children were treated with a standard first‐line therapy. We also describe the response rates to each of these standard therapies as well as to nonstandard therapies for IESS.

The subdivision of etiology plays a crucial role in the prognostication of IESS since a better prognosis is correlated with the absence of known etiology. ¹⁰ The understanding of etiological subtypes of IESS has expanded from the traditional known and unknown to six etiological categories based on the recent ILAE conceptual framework. ⁴ There are further subdivisions based on underlying genetic abnormalities and types of structural defects. ³ , ¹³ Apart from a few specific etiologies such as tuberous sclerosis complex ¹⁴ and Down syndrome, ¹⁵ the prognosis of IESS related to individual etiologies has been explored only minimally. Among those with acquired structural anomalies causing epileptic spasms, hypoxic‐ischemic injury, periventricular white matter injury, and hypoglycemia‐related brain injury are frequent etiologies, and they are associated with worse prognosis. They are reported in higher proportions in countries in the low‐income strata. For example, IESS due to hypoxic‐ischemic injury is reported in 58% in Bangladesh, 34% in India, 50% in Nepal, 32% in Pakistan, 28% in Myanmar, and 20% in Sri Lanka. ⁶ , ¹⁶ In our study, this proportion increased to 39%. These figures contrast with the 10% reported in patients in the United Kingdom Infantile Spasms Study (UKISS). ¹⁷ Due to its frequent association, potential biomarkers for early detection of those at risk for developing IESS in these perinatal insults, including periventricular leukomalacia ¹⁸ and hypoxic‐ischemic encephalopathy, ¹⁹ have been explored. In view of the variability in clinical presentation and poor response to therapy, further research is needed on IESS associated with hypoxic‐ischemic injury. Similar to hypoxic‐ischemic injury, hypoglycemia‐induced brain injury is another frequent cause of IESS in regions with poor maternal health status and perinatal care. It was reported as the single most frequent etiology, contributing to 35% of all cases of IESS in a recent study of 113 patients. ²⁰

Although it is important to understand the differences in the etiology of IESS in different regions, data from low‐income countries are limited. One reason for this is the lack of accurate data on etiology, as evident in our study, where a large group could not complete an evaluation of their etiology. A recent study from India could not identify etiology in 17%. ⁵ More studies on this aspect are needed since it creates a platform to lobby for improvement of healthcare provision, including improved perinatal care that may prevent the development of IESS, in resource‐limited countries.

The findings from SLISR highlight several aspects of management that need improvement. These include a median delay of 1.17 ± 1.33 months to commence therapy. This resulted in a large proportion (46%) being treated after a lag time of more than 28 days. Fourteen percent of the cohort received nonstandard therapy as the first therapeutic option, which contributed to poor outcomes. This may not necessarily indicate the choice made by the pediatric neurologist, since some babies were commenced on treatment by the pediatricians before referral to the neurologist. Some of the obvious reasons for not administering the readily available oral corticosteroids include poor nutritional status of the baby and concern about risk of ongoing infection, including some of the treatable congenital infections. Since both delays in diagnosis and inappropriate initial therapy may be directly associated with worse cognitive outcome, ¹⁰ our findings highlight the need for greater awareness and change in practice on early, appropriate therapy to mitigate time lost in achieving early spasm freedom.

Hypsarrhythmia was reported in 89% of our patients, a higher rate than reported in other series. This may reflect the poor interrater reliability in the interpretation. The overall rate of spasm cessation at the end of two weeks was approximately 60%. A further 18.5% showed a delayed response resulting in an overall 79% spasm freedom at the end of six weeks. This proportion aligns with mean response rates reported in most open‐label prospective studies, randomized clinical trials, and meta‐analyses for first‐line medications. ²¹ , ²² , ²³ , ²⁴ This corroborates the refractoriness of IESS, with approximately two‐thirds of infants achieving initial control at the end of six weeks. ⁷ Among Sri Lankan pediatric neurologists, the most frequent first‐line medication choice was oral prednisolone (64%), administered at a dose of 4–8 mg/kg/day (maximum of 60 mg/day). This choice is most likely influenced by its availability, easy dispensability, and low cost. ⁹ , ¹¹

The spasm control rate in our registry differed from that reported in the US registry, where a higher response rate of 55% was reported for ACTH compared to 39% for oral prednisolone. ²³ This may be related to larger doses as well as the different preparation of ACTH used in the US study. Due to its cost, the use of vigabatrin as a first‐line medication in low‐income countries is usually limited to children with tuberous sclerosis complex. The expected superior response with standard therapy for control of spasms ²³ as well as for electroclinical response was confirmed by the SLISR findings. The proportion who achieved electroclinical response was low, most likely due to a sizable proportion of children receiving standard therapy who did not undergo a repeat EEG. This participant attrition may have been due to successful control of spasms with the standard therapy. Our data suggest that neither early age of spasm onset (<3 months of age) nor delayed commencement of therapy (i.e., more than 14 days after spasm onset) were associated with worse control of epileptic spasms when assessed 14 days after therapy began. Additional studies are needed to confirm these two observations.

Our findings have several limitations. A large number of children failed to complete a posttreatment EEG, and many did not undergo a thorough evaluation for etiology. The latter is common in resource‐limited settings, mainly due to the limitations in availability and accessibility to the diagnostic studies needed to assess structural, metabolic, and genetic etiologies. As previously suggested by investigators from the South Asian region, this group of children can be categorized as “etiology incompletely evaluated.” ⁶ Nonavailability of consecutive samples across the participating sites may have bias towards inclusion of children who are more likely to respond to therapy. Despite these limitations, the relatively large sample size has enabled us to derive several useful observations on etiology and response to therapy and factors to be improved in the management of IESS.

Conclusion

Registries are a useful way to gather information about rare diseases. Data from the SLISR have helped us understand the characteristics of IESS and its response to therapy. While the data reveal etiologies more common to low‐income countries, questions remain about their true proportions due to difficulties in completing etiological investigations. Although a large majority of patients were treated with a recommended first‐line therapy, the treatment delay of more than 28 days in nearly half the patients was not optimal. The superior response to standard first‐line medications was confirmed, but neither an early age of spasm onset nor an increased time to treatment were associated with poor initial control of spasms.

Author Contributions

Jithangi Wanigasinghe: Conceptualization; funding acquisition; methodology; project administration; writing – original draft; writing – review and editing. Gemunu Hewawitharana: Data curation; project administration; validation. Pyara Ratnayake: Data curation; project administration; validation. Saraji Wijesekera: Data curation; project administration. Chathurika Weeraratne: Data curation; project administration. Ashan Jayawickrama: Data curation; formal analysis. Jayasanka Jayawardena: Formal analysis; methodology.

Conflicts of Interest

The authors declare no conflicts of interest.