Beyond Seizure Management in a Developmental Epileptic Encephalopathy

Abstract

Background and Objectives

SLC6A1-related neurodevelopmental disorder (SLC6A1-NDD) is emerging as a leading genetic cause of epilepsy, autism spectrum disorder (ASD), and NDD caused by monoallelic loss of function variants in SLC6A1 (solute carrier family 6 member 1) which encodes a gamma-aminobutyric acid (GABA) transporter (GAT1). Descriptions of SLC6A1-NDD have centered around epilepsy, but caregivers report other features that drive key aspects of the disease burden. This retrospective chart review and prospective natural history study (NHS) aims to highlight the nonseizure clinical features of SLC6A1-NDD.

Methods

Medical records of individuals with SLC6A1-NDD from University of Texas Southwestern (UTSW) and from the Children's Hospital of Philadelphia (CHOP) were reviewed. Participants had a confirmed or strongly suspected pathogenic variant in SLC6A1. Twenty of these individuals were also enrolled in an NHS at UTSW which included standardized scales and caregiver interviews.

Results

Sixty-three individuals were included (40 from UTSW, 23 from CHOP) with 31 males. Developmental delay was present in 97%, developmental regression in 41%, and 58% of those had an atypical regression. Of the total cohort, 60% have ASD. Maladaptive behaviors were present in a majority with aggression occurring in 57% of those. Twenty individuals completed standardized scales as part of the NHS at UTSW. The developmental coordination disorder questionnaire (DCDQ) demonstrated that 90% met criteria for probable developmental coordination disorder. Sleep problems were reported in 56%, but when assessed with the sleep disturbance scale for children (SDSC), all NHS participants scored >39, indicating clinically significant sleep problems. Total SDSC score positively correlated with behavior problems.

Discussion

Daily effects from SLC6A1-NDD cannot be described by epilepsy alone. Combining retrospective and prospective NHS data expands phenotypic and genotypic knowledge. We report the highest association with ASD and unusual developmental regression patterns. Maladaptive behaviors are frequent, commonly aggression. DCDQ and SDSC scores correlated with caregiver concerns but also found higher instances of movement and sleep problems, underrecognized by caregivers and providers. Often, odd movements defy categorization between stereotypies, tics, and epileptic movements. GABA homeostasis disruptions are integral to symptom formation but require better understanding.

Introduction

With an estimated incidence of 2.65 per 100,000 births (90% CI: 2.38–2.86¹). SLC6A1-related neurodevelopmental disorder (SLC6A1-NDD) is a leading genetic cause of epilepsy, autism spectrum disorder (ASD), and NDD.^2-7 Those affected have monoallelic loss of function variants in SLC6A1 (solute carrier family 6 member 1) which encodes a gamma-aminobutyric acid (GABA) transporter (GAT-1), responsible for reuptake of GABA from the synaptic cleft⁸ within the cerebellum, hippocampus, and basal ganglia.^9-13 Loss of GAT-1 expression leads to dysregulation of GABA and then can cause desensitization of GABA receptors. Communication between neurons and interneurons becomes impaired, especially in times of sustained neuronal activity.¹⁰

When GABA (an inhibitory neurotransmitter) sensitivity is decreased, neuronal network modulation is dysregulated. The fine control over network excitation and inhibition balance is diminished, and the patterns necessary for information processing, motor movement coordination, brain development, and synaptic plasticity are disrupted.^10-12,14 Disturbed network modulation generates spike-wave discharges that interrupt information flow through the thalamus, leading to behavioral arrest and absence seizures, a highly prevalent seizure semiology associated with SLC6A1-NDD.^9-13

The degree of GABA dysregulation is determined by the residual GAT-1 function associated with each genotype. Lower residual GAT-1 function correlates with more severe symptoms of SLC6A1-NDD.¹⁵ Studies are on-going, but residual GAT-1 activity appears affected by SLC6A1 variant location within the protein structure, particularly around sites important for GAT-1 trafficking to the cell membrane and interacting with GABA.¹⁵ To date, all pathogenic variants are classified loss-of-function (LOF). Most variants are missense, but nonsense and frameshift have also been identified.¹⁵ LOF may be complete or partial, and complete LOF is related to more severe symptoms of SLC6A1-NDD.^15,16

A retrospective review of 116 patients² and analysis of the SLC6A1-NDD cohort within the Simons Searchlight genetic NDD registry⁶ established that the core symptoms of SLC6A1-NDD are developmental delay (DD), epilepsy, and ASD with comorbidities of movement disorders, sleep challenges, and psychiatric problems. Absence seizures were identified as the primary seizure type, and the spectrum of disability was broad^{2,6,13-15,17,18}; contrasting with earlier studies that reported severe disability and epilepsy with myoclonic-atonic seizures as the primary seizure type.^5,7

Detailed reporting of the nonseizure features of SLC6A1-NDD has been limited. The aim of this study was to describe a cohort of individuals, both in retrospective chart review and preliminary data from a prospective Natural History Study (NHS), to expand on these drivers of disease burden and identify areas that need further study.

Method

Data Collection

Medical records of patients with SLC6A1-NDD were reviewed from the neurodevelopmental specialty clinic at the University of Texas Southwestern (UTSW) and from the pediatric neurology clinic at the Children's Hospital of Philadelphia (CHOP). In addition, data were collected through the prospective NHS of SLC6A1-NDD at UTSW. Demographics, neurodevelopmental histories, and frequency of comorbidities were reviewed, including ASD, seizures and semiology, problems with movement and coordination, development and regression, maladaptive behavior concerns, gastrointestinal difficulties, and sleep problems. Regression was categorized either “typical” or “atypical.” The characteristics of the latter defined as incongruent with regression associated with ASD, i.e., older than 3 years, isolated motor regression, or episodic regression. SLC6A1 genotypes were obtained from review of clinical genetic testing reports. Pathogenic, likely pathogenic, and variants of unknown significance (VUS) with clinical phenotype consistent with SLC6A1-NDD were included. A consistent clinical phenotype was determined by seizure semiology, developmental trajectory, and behavior profile expected for SLC6A1-NDD without a more likely explanation for symptoms. In addition, included VUSs were missense, a known mechanism for disease-causing variants, one individual had a documented pathogenic variant; however, genetic reports were not available at the time of this publication. The caregivers of NHS participants completed standardized questionnaires including the sleep disturbances scale for children (SDSC) and the developmental coordination disorder questionnaire (DCDQ) during their yearly visit beginning in 2020. Where applicable, the little DCDQ was administered. Clinical neuropsychological reports and school psychological evaluations were reviewed and summarized when available including standardized measures of ASD (Childhood Autism Rating Scale) and adaptive abilities (Adaptive Behavior Assessment Scale).

Outcome Measures

The following standardized clinical outcome rating scales were reviewed, where data were available.

The Childhood Autism Rating Scale, Second Edition (CARS-2), is a standardized measure used to assess the severity of ASD symptoms in children. The CARS-2 evaluates behavior, communication, and social interactions through observation and a structured caregiver interview. Scores range from 15 to 60 with scores 30 and over indicating ASD.¹⁹

The Adaptive Behavior Assessment Scale, Third Edition (ABAS-3), is a comprehensive tool designed to evaluate an individual's adaptive functioning in various life skills. It assesses an individual's ability to independently perform everyday activities required for personal and social sufficiency. The ABAS-3 covers domains such as communication, self-care, socialization, and motor skills. The ABAS generates a general adaptive composite (GAC) as well as individual domain scores. These fall on a normal curve with a mean of 100 and SD of 15. Higher scores indicate better adaptive skills.²⁰

The Developmental Coordination Disorder Questionnaire (DCDQ) is a caregiver report measure that evaluates motor coordination difficulties in children. The questionnaire assesses a child's proficiency in various motor activities, providing a comprehensive understanding of their coordination abilities. The little DCDQ is designed for 3- and 4-year-old children while the standard DCDQ is intended for ages 5 to 15 years old. The DCDQ has been validated as an appropriate tool to screen for developmental coordination disorder (DCD) with good diagnostic accuracy.²¹ Scores range from 15 to 75 with probable diagnosis stratified by age; a higher score indicates better performance, and a lower score indicates increased likelihood of DCD.²²

The Sleep Disturbances Scale for Children (SDSC) is a caregiver-report, clinical outcome measure designed to assess various sleep-related issues in children. This scale evaluates a range of sleep disturbances, including difficulties falling asleep, night waking, and parasomnias. Scores range from 26 to 130 with scores of 39 or greater indicating clinically significant sleep disturbance.²³

Statistical Analysis

Descriptive statistics, including mean and SD for continuous variables, frequency and percentages for categorical variables, were used to summarize clinical characteristics. Baseline clinical correlations were determined using the Wilcoxon rank sum test and Spearman rank-order correlation. Longitudinal trends were determined using the Wilcoxon signed-rank test. All statistical analyses were run on SAS 9.4²⁴ at a 0.05 significance level. Figures were generated in Tableau with the Infotopics extension²⁵ and R 4.4.0²⁶ with circlize²⁷ and trackViewer²⁸ packages. SLC6A1 protein modeling and variant mapping were built using Broad Institute's Genomics 2 Proteins portal.²⁹ Franklin by Genoox³⁰ was used to review variants and associated amino acid changes.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the UTSW and CHOP Institutional Review Boards. For the retrospective analysis, approval from UTSW and CHOP ethical standards committees to conduct this study was received. Participants in the NHS provided written informed consent.

Data Availability

Data not provided in the article because of space limitations may be shared (anonymized) at the request of any qualified investigator for purposes of replicating procedures and results.

Results

Cohort Demographics and Genetics

The combined retrospective clinical cohort of SLC6A1-NDD patients included 63 individuals (CHOP cohort n = 23, UTSW cohort n = 40) with 31 males (Table 1). History was collected from birth with the oldest participant 23 years old at the time of the latest clinical visit. Twenty of these individuals were also enrolled in a prospective NHS (Figure 1).

Table 1.

Demographics and Clinical Characteristics

Demographics: UTSW (including NHS) and CHOP, n (%)
Male:Female	31:32 (49:51)
Race	White	45 (71.4)
	Black/African American	3 (4.8)
	Asian	1 (1.6)
	More than 1 race	6 (9.5)
	Unknown/unavailable	8 (12.6)
Ethnicity	Hispanic/Latino	3 (4.8)
Demographics: UTSW NHS subset, n (%)
Male:Female	10:10 (50:50)
Race	White	17 (85)
	More than 1 race	1 (5)
	Unknown/unavailable	2 (10)
Age range (y) (median)	1.5–13.7 (7.6)
Clinical characteristics, n (%)
Developmental delay	61 (96.8)
Regression	26 (41.3)
Autism diagnosis	38 (60.3)
	Male:Female	18:20 (47:53)
	Level 1	1 (2.6)
	Level 2	9 (23.6)
	Level 3	16 (42.1)
	No level specified	12 (31.6)
Intellectual disability (n = 28)	24
	Mean IQ (n = 20)	70.2 ± 15.5
	Median IQ (n = 20)	69.5
ADHD diagnosis	18 (28.6)
Maladaptive behavior (n = 59)	46 (78.0)
	Tantrums/meltdowns	37 (80.4)
	Low frustration tolerance	16 (34.8)
	Aggression	26 (56.5)
	Self-injury	8 (17.4)
	Impulsivity	13 (28.3)
	Anxiety	13 (28.3)
	Selective mutism	3 (6.5)
Sleep disorder	35 (55.6)
Gastrointestinal disorder	39 (61.9)
Movement disorder symptoms	41 (65.1)
	Clumsy/uncoordinated	27 (65.9)
	Tremor	22 (53.7)
	Ataxiaa	6 (14.6)
	Abnormal gait	5 (12.2)
	Otherb	4 (9.8)
	Stereotypies (n = 39)	30 (78.9)

Abbreviation: ADHD = attention-deficit hyperactivity disorder.

Summary of sex, race, ethnicity, and clinical characteristics of participants, both the total cohort (63 participants) and the NHS (Natural History Study) group (20 participants). Where indicated, the number of participants (n) is lower than the total due to insufficient information in patient records.

^aAtaxia includes ataxia of gait as well as truncal ataxia.

^bOther movement symptoms include dystonia, tics, and chorea.

Figure 1. Natural History Study Participant Flowchart.

Of the 40 participants that were part of the UTSW cohort, 22 were enrolled in a Natural History Study (NHS). After enrollment, 2 individuals could not be contacted to schedule the first visit. After the first visit, 2 additional individuals were lost to follow-up. Seven families did not respond to schedule their third visit but were not yet considered lost to follow-up. As this NHS was ongoing, visits 4 and 5 were still being scheduled. CHOP = Children's Hospital of Philadelphia; DCDQ = developmental coordination disorder questionnaire; SDSC = sleep disturbance scale for children; UTSW = University of Texas Southwestern.

Genetic testing data were available for 62 participants with 37 variants classified pathogenic, 18 likely pathogenic, and 7 VUS (Table 2). Forty-three variants were missense and the remainder nonsense, frameshift, or intronic deletions (NM_003042.3 transcript). Using the hg38 assembly, variants were located between exons 3 and 16 with most variants located within exons 3 to 10 (Figure 2B). There was not clear correlation between variant location and categorical clinical characteristics (Figure 2C).

Table 2.

Participant Genetic Variants in SLC6A1

Genotype	Variant frequency	Protein change	Variant type	Pathogenicity	Inheritance
c.79C>T	1	p.(P27S)	Missense	VUS	Unknown
c.104dupA	1	p.(K36Efs*171)	Nonsense	Likely pathogenic	De novo
c.130C>T	2	p.(R44W)	Missense	Pathogenic	De novo
c.140G>A	1	p.(W47*)	Nonsense	Likely pathogenic	Unknown
c.154G>A	1	p.(D52N)	Missense	Pathogenic	De novo
c.207G>T	1	p.(R69S)	Missense	Likely pathogenic	De novo
c.235G>A	2	p.(G79R)	Missense	Likely pathogenic	Inherited from mosaic parent x2
c.302A>G	1	p.(Q101G)	Missense	Pathogenic	Unknown
c.331G>A	1	p.(G111R)	Missense	Pathogenic	De novo
c.339delA	1	p.(V115Yfs*22)	Nonsense	Pathogenic	Unknown
c.340G>A	2	p.(G114R)	Missense	VUS	Unknown, but inheritance likely
c.370+2del	1	N/A	Noncoding	Likely pathogenic	Unknown
c.373G>A	1	p.(V125M)	Missense	Pathogenic	Inherited (parent affected)
c.418T>C	1	p.(Y140H)	Missense	Pathogenic	Unknown
c.447C>G	1	p.(Y149*)	Nonsense	Likely pathogenic	Unknown
c.469A>G	1	p.(N166D)	Missense	Likely pathogenic	De novo
c.502del	1	p.(W168Gfs*18)	Nonsense	Pathogenic	Unknown
c.583C>T	1	p.(R195C)	Missense	VUS	Unknown
c.640_658del	2	p.(L214Sfs*32)	Frameshift	Pathogenic	Unknown
c.684_689delTATCTG	1	p.(C228_I229del)	In-frame microdeletion	Likely pathogenic	De novo
c.741del	1	p.(Y248Tfs*4)	Frameshift	Pathogenic	De novo
c.863C>T	4	p.(A288V)	Missense	Pathogenic	Maternal x 1, de novo x 3
c.884C>T	1	p.(S295L)	Missense	Likely pathogenic	De novo
c.889G>A	3	p.(G297R)	Missense	Pathogenic	De novo x 3
c.913G>A	2	p.(A305T)	Missense	Pathogenic	Unknown
c.919G>A	3	p.(G307R)	Missense	Pathogenic	De novo
exon 9–10 del	1	p.(V284*)	Nonsense	Pathogenic	Unknown
c.980A>G	1	p.(N327S)	Missense	Likely pathogenic	De novo
c.982del	1	p.(S328Rfs*45)	Frameshift	Likely pathogenic	De novo
c.991A>G	2	p.(S331G)	Missense	Likely pathogenic	De novo
c.1000G>T	3	p.(A334S)	Missense	Vus	Unknown, but inheritance likely
c.1070C>T	2	p.(A357V)	Missense	Pathogenic	Possible paternal mosaic x 1, de novo x 1
c.1084G>A	4	p.(G362R)	Missense	Pathogenic	Inherited from mosaic parent x 3, de novo x 1
c.1120C>T	1	p.(Q374*)	Nonsense	Pathogenic	De novo
c.1157T>C	2	p.(F386S)	Missense	Pathogenic	De novo x 1, unknown x 1
c.1192-2A>C	1	N/A	Splicing	Likely pathogenic	Inherited (parent low level mosaic [2–3%])
c.1205A>G	1	p.(E402G)	Missense	Likely pathogenic	De novo
c.1222delC	1	p.(L408Wfs*26)	Nonsense	Pathogenic	Unknown
c.1426+1G>A	1	N/A	Noncoding	Likely pathogenic	Unknown
c.1527+2T>C	1	N/A	Noncoding	Pathogenic	De novo
c.1528-1G>A	1	N/A	Noncoding	Pathogenic	De novo
c.1580_1581del	1	p.(Y527Cfs*58)	Frameshift	Likely pathogenic	De novo

Figure 2. Demonstration of Genetic Variants.

(A) GAT1 protein structure with representation of participant variants. (B) Variant number and type by exon: chromosome 3: 11017059 - 11039247; SLC6A1 exons 3-16. (C) Illustrative relationship between participant symptoms and participant variants. (A) Model of the GAT-1 protein encoded by the SLC6A1 gene with variants mapped and colored by pathogenicity. Truncating/nonsense variants are not mapped due to their effect on protein structure. (B) Demonstrates variant location along chromosome 3 using hg38 assembly. Missense variants were most common. “Other variants” include nonsense, frameshift, and noncoding. Variants are designated by amino acid (p.) nomenclature unless the variant is noncoding in which case it is denoted by nucleotide nomenclature (c.). (C) Indicates collected clinical characteristics and their association with participant variants which are identified and colored per exon as in panel B. From this the clinical characteristics in this cohort do not appear to cluster in any one part of the gene.

There was 1 participant without complete genetic data but with a pathogenic SLC6A1 variant documented in the medical record. Thirty-seven variants were classified as ‘Pathogenic’, eighteen classified as ’Likely Pathogenic’, and seven as ‘VUS’ (Variant of Uncertain Significance). Participants with VUSs were included when clinical phenotype was consistent with SLC6A1-NDD. Included VUSs were missense, a known mechanism of pathogenicity. A consistent clinical phenotype was determined by seizure semiology, developmental trajectory, and behavior profile expected for SLC6A1-NDD without a more likely explanation for symptoms. Where there was conflict in pathogenicity classification, the higher certainty classification is reported.  There are noted cases where inheritance is unknown but suspected, for example siblings with the same variant, but  parental testing was not available available. In one case, paternal mosaicism is suspected due to the subject’s half-sibling having the same variant, but the variant was not found on paternal testing.

Cognitive and Behavioral Phenotypes

Records of 63 participants were reviewed and revealed delay in 1 or more developmental domain in 61 (96.8%) individuals and developmental regression in 26 (41.3%) individuals. Within the NHS cohort, 20% (4/20) had language regression and 4 of 20 had motor regression, as reported by caregivers. In this group of 8 of 20, the mean age of regression was 32.5 months old with a range of 12–60 months. Six of these 8 (75%) recovered lost skills, and 2 did not regain lost language skills. Developmental trajectories of the 20 NHS cohort participants are fully described in a previous publication.³¹

Within the full cohort of n = 63, the pattern of developmental regression diverged in some individuals from what could be expected in regression associated with ASD; categorized “atypical regression” (Figure 3). Of the 26 individuals with regression, 3 (11.5%) did not have an ASD diagnosis. Fifteen (57.7%) had an atypical regression. The oldest age of regression was documented to have onset between 18 and 19 years old. This individual at baseline was able to express his needs, could perform most hygiene tasks independently, and had no movement disorder symptoms. With developmental regression, he had a significant decline in his ability to communicate effectively, perform self-care tasks, developed a wide-based gait and abnormal hand movements. He had a history of well-controlled epilepsy without a clear change in his seizure burden or EEG during regression.

Figure 3. Development and Regression Trends.

(A) Development and regression Sankey. (B) Development and regression cross-tabulation. Demonstration of regression and the relationship to autism spectrum disorder (ASD) and developmental delay (DD) diagnoses within SLC6A1-NDD. In panel A, flow is arranged based on the presence or absence of developmental delay, autism, regression, atypical regression, and typical regression. Panel B is a cross-tabulation of the same information, better illustrating that 3 individuals have not been diagnosed with ASD but did experience a regression. One individual did not have ASD or DD but did have an atypical regression. Of the 26 individuals with a developmental regression, a little over half had a regression defined as atypical.

Where the presence or absence of an intellectual disability (ID) diagnosis was explicitly stated in records (n = 28), 24 had a documented ID diagnosis. Based on available standardized testing reports (n = 20), the mean IQ score was 70.2 (±15.5), ranging from 44 (severe ID) to 99 (normal). A score of 70 or lower (2 standard deviations below the mean) is considered consistent with ID.³²

Thirty-eight of the 63 participants had a documented diagnosis of ASD (60.3%). Of these, 16 were assigned a level 3 ASD diagnosis, 9 level 2, and 1 level 1. A designated level was not available in 12 participants. ASD diagnosis was evenly split between male (n = 18) and female (n = 20) with similar severity; about 40% with an ASD diagnosis were designated level 3 in each sex. The available CARS-2 scores (n = 17) ranged from 23 to 58 with an average of 38.9 (±9.7), consistent with severe ASD. Fourteen CARS-2 scales were consistent with a diagnosis of ASD. Of those with available results for the ABAS-3 (n = 10), the mean GAC was 67.9 (±14.7, range: 41–82), >2 standard deviations below the standard population mean, indicating significant deficits in adaptive functioning.²⁰

Maladaptive behavior was reported in 46 of 59 participant records (78%). Of these 46, the large majority were documented to have tantrums/outbursts that were abnormal for age and disrupted the daily functioning of the individual and/or caregiver (37/46, 80.4%). While independent symptoms, low frustration tolerance (16/46, 34.8%), impulsivity (13/46, 28.3%), and anxiety (13/46, 28.3%) often occurred alongside reports of outbursts. Aggression was a concern in about half (26/46, 56.5%). Self-injury was less common (8/46, 17.4%). Attention deficit hyperactivity disorder was present in 18 of 63 (28.6%).

Movement Disorder Phenotypes

Caregivers observe that individuals with SLC6A1-NDD move differently than peers, describing poor coordination, problems with balance, twisting and posturing movements of fingers, clumsiness leading to injuries, tremulous movements, teeth grinding, body rocking, and tensing postures. In the combined cohort record review, 41 of 63 (65.1%) documented problems with movement. Difficulty with coordination (e.g., balance issues and general clumsiness) was the most common concern (27/63, 65.9%).

The DCDQ and little DCDQ were administered to the NHS cohort (Figure 4). The DCDQ was completed by the caregivers of 17 participants (10M:7F, age range from 5 to 15 years) with 15 meeting the threshold for diagnosis of a DCD. Lower scores on the DCDQ and little DCDQ indicate worse coordination performance. The mean total score for the DCDQ was 29.1 (±9.6). Cutoff scores are stratified by age and range between 46 and 57 (falling below these values indicates DCD). The mean score for control during movement was 12.2 (±4.3, possible maximum of 30), fine motor was 6.1 (±2.2, out of 20), and general coordination was 10.9 (±4.1, out of 25). While all scores were low when compared with typical, the level of ASD diagnosis had a strong negative correlation with low total DCDQ scores (correlation = −0.7, p value = 0.002). The little DCDQ (LDCDQ) was completed by the caregivers of 10 participants (3M:7F, age range 1–5 years). The mean total score was 38.2 (±19.1) (scores below 68 indicate probable DCD) with a mean gross motor score of 23.4 (±12.2) and mean fine motor score of 14.8 (±7.9). Each subdomain in the LDCDQ has a maximum of 25. Of the 10 LDCDQ participants, 3 of these were not old enough in subsequent visits to complete the DCDQ. Two of these 4 were classified “probable DCD” based on their LDCDQ scores. In total, 18 of 20 (90%) participants met criteria for probable DCD.

Figure 4. Longitudinal Trends for DCDQ; Comparison With Participant Characteristics.

(A) DCDQ and little DCDQ longitudinal trends. (B) Little DCDQ categorical traits heatmap. (C) DCDQ categorical traits heatmap. (A) Longitudinal total scores of the LDCDQ and the DCDQ. As participants aged, some were able to cross from the LDCDQ assessment to the DCDQ. On the DCDQ, the score at which “probable DCD” is determined is stratified by age. The dashed lines indicate the level below which a child is considered to have probable DCD. (B and C) demonstrate correlations between LDCDQ and DCDQ scores with participant characteristics. Highlighted scores indicate strong correlations between total scores or subdomain scores and participant variables. In figures B and C, the level of ASD diagnosis had a strong negative correlation with both LDCDQ and DCDQ total score (−0.6, p value = 0.053; −0.7, p value = 0.002, respectively). Otherwise stated, the more severe ASD diagnosis correlated with more severe coordination problems. In figure B, the authors also highlight that onset of seizures at an older age was moderately correlated (0.5, p value = 0.167) with a better fine motor score on the LDCDQ. In panels B and C, positive values (red) indicate a direct correlation, while negative values (blue) indicate an inverse relationship. Values indicate the rho value derived from Spearman rank correlation. ASM = antiseizure medication; DCDQ = developmental coordination disorder questionnaire; LDCDQ = little DCDQ.

Sleep Disorder Phenotypes

Disrupted sleep patterns and insomnia negatively affect quality of life.³³ In the combined cohort records, sleep disorders were reported in 35 of 63 (55.6%). From the NHS cohort, caregivers reported 11 of 20 (55%) had difficulty falling asleep, 8 of 20 (40%) had difficulty with sleep maintenance, 4 of 20 (20%) had restless sleep, 3 of 20 (15%) had early morning waking, and 4 of 20 (20%) reported no sleep problems.

The SDSC was administered to all participants in the NHS (n = 20) (Figure 5). All participants had a score >39 (with a mean total score of 65.2 ± 17.0), indicating clinically significant sleep problems.²³ Longitudinally, there was improvement with an average reduction of 8.2 points over 2 years (p = 0.013). When comparing with caregiver sleep concerns, disruptive sleep (lacking continuous, uninterrupted sleep) was positively correlated with SDSC subdomains Disorders of Arousal (DA) (47 ± 0 vs 61.3 ± 20.6, p value = 0.03) and Sleep-Wake Transition Disorders (SWTD) (26.7 ± 40.1 vs 71.6 ± 36.8, p value = 0.02). Disorders of Initiating and Maintaining Sleep (DIMS) and Disorders of Excessive Somnolence (DOES) both positively correlate with a diagnosis of ASD (57.8 ± 8.4 vs 72.8 ± 22.5 [p value = 0.04], and 49.3 ± 7.7 vs 58.6 ± 18.9 [p value = 0.04], respectively). The DOES subdomain had a moderate correlation coefficient of 0.5 to level of ASD (p value = 0.04). The DIMS also positively correlated with overall clinical severity (73.7 ± 19.7 severe vs 53.5 ± 7.4 mild, p value = 0.01). Finally, the total SDSC positively correlated with the presence of behavior problems (69.2 ± 16.5, p value = 0.005). There was no correlation between the SDSC total score and any subdomain score with age.

Figure 5. Longitudinal Trends for SDSC; Comparison With Participant Characteristics.

(A) SDSC longitudinal trends. (B) SDSC categorical traits heatmap. In (A), the trend of sleep disturbances across study visits is captured with overall minor improvement in sleep. Average scores remain stable for visits 2–3 but begin to fall in visit 4 (lower scores indicate fewer sleep disturbances). As data collection was ongoing for visits 4 and 5 during the writing of this article, we are unable to capture the final trend. Numbered circles indicate participant ID. (B) demonstrates the subdomains of the SDSC and their correlation with participant characteristics. Highlighted scores indicate strong correlations between total scores or subdomain scores and participant variable. The SDSC total scores were moderately to strongly correlated with the number of sleep medications an individual was prescribed (0.6, p value = 0.004) and number of behavior medications prescribed (0.6, p value = 0.003). SHY and age of seizure onset were negatively correlated (−0.5, p value = 0.021). In addition, the DOES subdomain correlated with severity of autism diagnosis (0.5, p value = 0.04). In panel B, positive values (red) indicate a direct correlation, while negative values (blue) indicate an inverse relationship. Values indicate the rho value derived from Spearman rank correlation. ASM = antiseizure medication; DA = disorders of arousal; DCDQ = developmental coordination disorder questionnaire; DIMS = disorders of initiating and maintaining sleep; DOES = disorders of excessive somnolence; SBD = sleep breathing disorder; SDSC = sleep disturbances scale for children; SHY = sleep hyperhidrosis.

Gastrointestinal Phenotypes

Gastrointestinal (GI) issues ranged from reflux and restricted eating to constipation and severe food allergies.³³ Of 63 participant records, 39 (61.9%) had a GI disorder and in the NHS cohort, caregivers reported 5 of 20 (25%) had a restricted diet.

Discussion

Epilepsy is only part of the individual and caregiver journeys with SLC6A1-NDD. By combining prospective NHS data with retrospective review of patient data gathered from 2 major medical centers, we aim to expand the knowledge, phenotypes, and genotypes of SLC6A1-NDD.

Understanding of developmental and epileptic encephalopathies (DEEs) is evolving. In 2017, the ILAE revised the term epileptic encephalopathy to DEE to recognize that the underlying pathology of disease contributes to developmental differences and that it is not seizure burden alone that leads to alterations in development.^34,35 Consistent with previous studies of SLC6A1-NDD, DD was ubiquitous in the combined cohort (96.8%, 61/63). Developmental regression has been described in SLC6A1-NDD³¹ and occurs in a subset of individuals with ASD from any cause.³⁶ Regression in ASD does not have a standard definition, but generally accepted as sudden or gradual loss of skills (language and social) between 15 and 30 months old (average onset at 19.8 months).^36,37 A meta-analysis of more than 33,000 individuals found a 30% global prevalence of regression in ASD.³⁶ Approximately 41% (26/63) of this SLC6A1-NDD cohort have a history of regression in some area of development, higher than the global prevalence of regression in ASD. In the NHS cohort (n = 20), regression was evenly split with 20% losing language skills (expected in ASD), but another 20% had motor skill regression, which is atypical. Of those with regression in our combined cohort, about 58% (15/26) had a regression that could be considered atypical, i.e., >3 years old, isolated motor regression, or an episodic regression (Figure 2). Most participants with any kind of regression also had ASD, but there were 3 participants with a regression without ASD (1 with a typical regression and 2 with an atypical regression). The mean age of regression onset in SLC6A1-NDD previously reported was 32–33 months,³¹ (older than the global mean age at onset for ASD). A shift in cognitive abilities often accompanies DEEs in childhood, sometimes with seizures and sometimes independent of seizures due to regression or stagnation.^35,38 Decline can continue into adulthood, despite seizure control, as evidenced by Dravet syndrome (particularly those with an SCN1A variant),^35,38 STXBP1 (where 26% in a recent study of 32 adults had regression beyond early childhood),³⁹ and by the patient reported in this study with a functional regression at 18 years old. It was beyond the scope of this study to explore whether periods of regression had any correlation with seizure onset or worsening.

In the full cohort of 63 participants, ASD was reported in 60.3% of which 42.1% had level 3 ASD with only a single level 1 ASD. Previously, the highest ASD prevalence in SLC6A1-NDD reported was 30%.^2,6 We had many participants without a reported level (12/38). Given that a level 2 or 3 diagnosis indicates more support needs, these levels could be subject to ascertainment bias and more frequently reported in the medical record, skewing our numbers toward more severe. Even at 60.3%, ASD may still be underreported. ASD diagnoses may not have been explicitly stated in the clinical reports reviewed (especially if support needs are low), and there were several children who had “autistic traits” but were not yet evaluated, in part due to difficulty accessing ASD evaluations.

Deficits in adaptive skills were documented, especially in those with ASD, as expected,. However, even without an ASD diagnosis, adaptive skills were disrupted, as illustrated by 2 participants who did not meet ASD criteria but had “borderline” and “below average” GAC scores on the ABAS-3. Low adaptive skills in this group support that ID diagnoses are likely in SLC6A1-NDD,⁶ but when examining the IQ scores available in 20 participants, there was a large SD of ±15.5 from the mean of 70.2, ranging from severe ID to normal intelligence. In fact, 50% of available IQ scores were above 70, consistent with normal intelligence, suggesting a broader spectrum of cognitive ability. Likely, previous studies were drawing from a pool of more severe phenotypes (i.e., severe seizures, severely delayed), but expanded genetic testing has increased phenotypic variability available. Conversely, available IQ testing may have been more commonly completed when presentations were less clear between normal intelligence and mild ID.

Maladaptive behavior was described in 46 of 59 records (78%). Aggression was particularly troublesome for caregivers and was present in 56.5% (26/46) of those with any maladaptive behavior, and present in 41% (26/63) of our total cohort. In children with ASD from any cause, aggression is present in about 25%.⁴⁰ Caregivers frequently described the behavior as being unpredictable. A previous study found oppositional and attention disorders were reported by 92–100% of caregivers using the Child Behavior Checklist (for 25 individuals with SLC6A1-NDD)⁶. The aggression subdomain was of clinical concern in 8% of 1.5 to 5 year olds (13 individuals) increasing to 33% in 6 to 18 year olds (12 individuals). Further analyses were outside the scope of this study but should be included when developing standardized studies of the behavioral phenotype in this population.

Developmental coordination disorder (DCD) describes poor motor proficiency that prevents the child from adequately building the complex skills that ultimately lead to performing activities of daily living.^32,41,42 Children with DCD who are otherwise neurotypical experience frustration and social isolation because they are unable to keep up with their peers in both academics (due to writing difficulties) and physical activities, leading to motivational and behavioral issues.⁴² Lower IQ and DD are associated with motor performance challenges.⁴¹ Children with ASD specifically have atypical patterns in gait and reach, regardless of cognitive level.⁴³ Within our record review, caregivers reported general movement problems in 65% (41/63) of participants, usually discoordination. They had trouble with fine motor skills, may fatigue during sports, or issues may be more severe including dystonia, tremors, and abnormal gait. This is higher than a previously reported 48% that included tremor, ataxia, or dystonic movement⁶ and is consistent with movement disorder symptoms in other DEEs.³⁴ Another study using the DCDQ in children with epilepsy found that fine motor skills (particularly handwriting) were most affected, followed by the ability to learn new skills.⁴⁴ Our participants displayed similar difficulties, with the fine motor domain most challenging overall, but the worst performing single item was the ability to learn new skills. Our criteria may be more inclusive for defining “problems with movement,” but in the NHS cohort, 90% (18/20) were classified “probable DCD” by DCDQ and LDCDQ, suggesting that movement problems may be undercounted.

As expected, participants in the NHS with ASD were more likely to have coordination issues. Level of ASD was found to negatively correlate with severity of DCD, i.e., a participant with level 3 ASD would have a lower (more severe) score on the DCDQ. However, even the participants who did not have ASD had a lower DCDQ score than expected for age.

Among the full cohort (n = 63), stereotypies were frequent, not always associated with ASD, and difficult to delineate from seizures. Early events of eye fluttering (usually with staring) and drops trigger seizure evaluation but may not have an EEG correlate, and might be more representative of a tic, stereotypy, or underlying coordination issues. These events sometimes respond to antiseizure medications even without EEG correlate. A more specific analysis of which medications are effective and why is outside the scope of this study.

In the NHS cohort (n = 20), the total SDSC scores were correlated with behavior problems, suggesting a relationship between disrupted sleep and increased behavior concerns. The SDSC subdomains relating to initiating and maintaining sleep and excessive somnolence were both strongly correlated with a diagnosis of ASD. Sleep disturbances are a common problem in ASD with up to 95% of caregivers reporting sleep problems⁴⁵ and severity of core symptoms seem to have a bidirectional relationship with sleep disturbance severity,⁴⁶ consistent with our study. In a study also using the SDSC to evaluate sleep disorders in children with ASD vs typically developing children, the SWTD (Sleep-Wake Transition Disorders) and the DIMS (Disorder of Initiating and Maintaining Sleep) subdomains were elevated (similar to our study), while in typically developing children, the Sleep Breathing Disorder (SBD) and the Sleep Hyperhidrosis (SHY) domains were elevated.⁴⁶ All our NHS participants had clinically significant sleep problems on the SDSC, even when an ASD diagnosis was not present. However, the previously mentioned study reported that lower adaptive skills were also correlated with increased sleep problems.⁴⁶ In SLC6A1-NDD, consistently low adaptive skills, consistently poor sleep, and consistently difficult behavior are potentially interconnected. Caregivers reported sleep challenges in 80% (16/20) of the NHS cohort and about 56% (35/63) of the combined cohort, but 100% of NHS participants demonstrated clinically significant sleep disorders on the SDSC, suggesting that caregivers may not recognize symptoms of sleep disorders or they are not routinely asked. Over the course of the NHS, the SDSC scores improved, although there was no correlation between age and SDSC scores. Individuals may have developed more mature sleep patterns or behavioral and medication interventions were put in place as they continued in standard clinical care outside the NHS. Evaluating for age-related sleep changes vs sleep interventions was outside the scope of this study and limited by ongoing NHS data collection.

There are several limitations with this study. Our NHS cohort was small, so our use of standardized scales was constrained. Since this a preliminary report of NHS data, trend analysis is limited. Outside of the NHS cohort, data gathered were based on reports and chart review, whereas standardized evaluations of areas such as behavior and development would have been preferred. The nature of caregiver-reported information is subject to recall bias. Results are also limited by the availability of participants to receive diagnosis, care, and travel to the institutions.

This observational study that incorporates both prospective and retrospective data collection is valuable as a hypothesis-generating tool. DEEs have many commonalities, but standard questioning and reporting often lacks the nuance needed to fully describe SLC6A1-NDD. For example, ASD can be a missed or delayed diagnosis in SLC6A1-NDD because social motivations may be largely preserved, but other abnormalities in social communication and behavior persist. Our understanding of SLC6A1-NDD and potentially DEEs in general, would benefit from dedicated characterization of regression to better understand whether there are triggers (e.g., seizure onset, illness, and EEG changes), how many truly regain skills, and the differences in regression occurring in younger vs older individuals.

Behavior is frequently reported by caregivers as challenging but is not well quantified. Comparing against the larger ASD population, children with SLC6A1-NDD seem to be more prone to aggression than other NDDs, making their behavior uniquely challenging. We need to better understand how the development, regression patterns, and behavior patterns in SCL6A1-NDD are unique to better tailor evaluations and interventions for these individuals. Tools for treating seizures in SLC6A1-NDD have continued to expand with investigations into acetazolamide and 4-phenylbutyrate showing promise.^47,48 However, the same cannot yet be said about novel approaches to behavior concerns.

Although there were no formal qualitative insights gathered from caregivers within this study, the measures selected were based on review of the SLC6A1-NDD draft conceptual model³³ and input from the SLC6A1 Connect network of caregivers and collaborators. Sleep and motor coordination problems were identified as key effects of disease burden. Items on the SDSC and DCDQ capture sleep and discoordination symptoms that reflect caregiver concerns. For example, in the NHS cohort (n = 20), caregivers describe disruptive sleep, which then correlated with DA (Disorders of Arousal) and SWTD subdomains on the SDSC. Beyond our own study, sleep scales have previously demonstrated accuracy in capturing sleep disorders.⁴⁶ The DCDQ is meant as a screening tool and does not replace trained evaluation, but with a sensitivity of 0.87 and specificity of 0.83 for DCD, as well as its already-established use in many countries, it remains a valuable tool for future studies.²¹

In conclusion, this study expands the understanding of SLC6A1-NDD and expectations for patients, providing an in-depth examination of caregiver reports, clinician evaluation, and standardized scales. The SDSC and DCDQ capture sleep and coordination problems, respectively, that are reported by parents and support the retrospective review of similar problems. We provide actionable areas to evaluate and intervene, drawing attention to areas that are frequently underreported in DEEs. While expanding the SLC6A1-NDD phenotype, we also stress the importance of sharing knowledge and developing robust NHS in future rare disease studies to provide solid groundwork for clinical trials.

Glossary

ASD

autism spectrum disorder

CHOP

Children's Hospital of Philadelphia

DCD

developmental coordination disorder

DCDQ

developmental coordination disorder questionnaire

GABA

gamma-aminobutyric acid

GAC

general adaptive composite

GI

gastrointestinal

LOF

loss-of-function

SDSC

sleep disturbance scale for children

SLC6A1-NDD

SLC6A1-related neurodevelopmental disorder

UTSW

University of Texas Southwestern

Author Contributions

M.E. Varnet: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. H. Dahshi: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. K. Goodspeed: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. P.P. Mcdonnell: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. L. Lusk Stripe: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. D.M. Armstrong: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data.

Study Funding

The natural history study was funded by the patient advocacy group, SLC6A1 Connect.

Disclosure

M. Varnet is currently a full-time faculty member at the University of Colorado. The work presented here was completed during her previous employment as a resident at the University of Texas Southwestern Medical Center; manuscript completion and revision was completed after her transition to University of Colorado. H. Dahshi reports no disclosures relevant to the manuscript. K. Goodspeed is currently a full-time employee with Ultragenyx Inc. The work presented here was completed during her previous employment as faculty with the University of Texas Southwestern Medical Center and is not related to her work affiliated with Ultragenyx Inc. P.P. McDonnell, L. Lusk Stripe, and D.M. Armstrong report no disclosures relevant to the manuscript. Go to Neurology.org/NG for full disclosures.