Clinical signatures of SYNGAP1-related disorders through data integration

Abstract

Purpose:

SYNGAP1 is a genetic neurodevelopmental disorder characterized by generalized epilepsy, autism, and intellectual disability. Despite a comparatively high prevalence, the longitudinal landscape remains relatively unexplored, and complete characterization is essential for clinical trial readiness.

Methods:

We combined electronic medical record data (n=158) with insurance claims data (n=246) to evaluate longitudinal progression of symptoms.

Results:

Phenotypes associated with SYNGAP1 included behavioral abnormalities (Odds ratio (OR) 12.35, 95% CI 9.21–16.78), generalized-onset seizures (OR 1.56, CI 1.20–2.02), autism (OR 12.23, CI 9.29–16.24), and a developmental profile with prominent deficits in verbal skill acquisition. Several clinical features showed distinct age-related patterns, such as a more than five-fold risk of autistic behavior emerging between 27 and 30 months. Generalized-onset seizures were significantly increased (OR 4.05, CI 2.02–7.59) after 3 years of age and persisted over time. Valproic acid and clobazam were commonly used for epilepsy treatment, while risperidone, aripiprazole, and guanfacine were commonly used for behavior management. Valproate and lamotrigine were more effective at reducing seizure frequencies or maintaining seizure freedom than other anti-seizure medications.

Conclusion:

We delineated the seizure, developmental, and behavioral trajectories in SYNGAP1-related disorders, to improve diagnosis, prognosis, and clinical care, as well as facilitating clinical trial readiness.

INTRODUCTION

Disease-causing variants in SYNGAP1 (HGNC:11497) are among the most common monogenic etiologies for generalized epilepsy and non-syndromic intellectual disability, with an estimated prevalence of 6 per 100,000.¹ Since its first description in human epilepsy in 2009,² pathogenic variants SYNGAP1 are now recognized as a common cause of developmental and epileptic encephalopathies (DEEs) in addition to several distinct epilepsy syndromes, such as epilepsy with myoclonic atonic seizures (EMAtS) and epilepsy with eyelid myoclonia (EEM).^3–8 However, the full range of clinical presentations in individuals with SYNGAP1-related disorders (SYNGAP1-RD) and the trajectories of symptoms over time have been insufficiently delineated. This knowledge gap represents a major impediment for precision medicine trials as the longitudinal course of SYNGAP1-RD is incompletely understood.

SYNGAP1 encodes a key regulatory protein of the post-synaptic density (PSD) at the interface of NMDA receptors and the downstream signaling apparatus, which is essential for synaptic plasticity.^9–11 The main disease mechanism is haploinsufficiency, and the majority (~74%) of individuals with SYNGAP1-RD have protein-truncating variants.¹² SYNGAP1 represents a focused target for novel precision medicine approaches including the development of antisense oligonucleotides (ASOs).^13–15

Clinical research on SYNGAP1-RD has largely been performed through small cohort studies with focused phenotypic descriptions, and only limited genotype-phenotype associations have been described to date.^6,7,12 Furthermore, especially in early childhood prior to epilepsy onset in SYNGAP1-RD, the phenotypic picture is frequently non-specific and the presentation of individuals with SYNGAP1-RD may resemble other neurodevelopmental conditions such as Rett Syndrome and Angelman Syndrome.^16,17 An early molecular diagnosis has implications in the clinical management, including early intervention and tailored care for improving long-term developmental outcomes. Accordingly, early symptom recognition and clear delineation of longitudinal trajectories in SYNGAP1-RD remain critical.

Here, we characterize phenotypic features, including behavioral abnormalities and seizure types, in individuals with SYNGAP1-RD across the age-span using real-world data captured from various large-scale healthcare resources. To illustrate the unique features of this disorder, we compare the clinical features in SYNGAP1-RD with a broader population of individuals with epilepsy and neurodevelopmental disorders, outlining the distinctive longitudinal landscape of epilepsy and developmental trajectories of this rare disorder.

METHODS & PARTICIPANTS

Inclusion of individuals with SYNGAP1-RD from various healthcare resources

We defined two broad cohorts of individuals with SYNGAP1-RD (Table 1)—those obtained from healthcare insurance claims data (n=246) and those extracted from electronic medical records (EMR) data (n=158). We identified individuals with SYNGAP1-RD through large-scale healthcare claims data (n=246) using the International Classification of Diseases,¹⁸ Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis code for SYNGAP1 (F78.A1). Inclusion criteria were: (1) SYNGAP1 F78.A1 diagnosis codes across at least two distinct encounters, or (2) an epilepsy, developmental delay, or intellectual disability diagnosis in addition to a SYNGAP1 F78.A1 diagnosis. Only individuals under the age of 25 years were included, as the recency of the ICD10-code limited duration of clinical histories captured in the database for older individuals. To create the electronic medical records cohort (n=158), we included all individuals in the Citizen Health Natural History Registry and individuals with SYNGAP1-RD who received care at Children’s Hospital of Philadelphia (CHOP). Individuals followed at CHOP were identified via free text EMR search for “SYNGAP1” followed by manual chart review. Individuals with data in both the Citizen Health and CHOP datasets were identified by shared variant, sex, and age, and data from any duplicates were combined. To account for cohort overlap, we analyzed healthcare claims data separately from electronic medical record data.

Table 1.

Cohort of individuals with SYNGAP1-RD.

	Healthcare claims	Electronic medical records
Cohorts (ICD10-CM)a  SYNGAP1-related disorders (F78A1)  Broad epilepsy cohort (G40+)  Angelman Syndrome (Q9351)  Rett Syndrome (F842)	n=246 individuals n=680,076 individuals n=3,697 individuals n=5,489 individuals	n=158 individuals -- -- --
Demographics of SYNGAP1 cohort  Sex  Race  Ethnicity  Median age at assessment (IQR)  Median observation time (IQR)  Median age of diagnosis (IQR)  Total patient-years  Variant spectrum	140 male, 106 female unavailable unavailable 6.53 years (1.5 – 12.5 years) 6 years (4.8 – 6.3 years) 11.2 years (5.6 – 16.5 years)b 1,321 years --	80 male, 78 female 11 white, 3 Asian, 1 multipled 5 Hispanic, 11 non-Hispanicd 5.6 years (3.8 – 10.0 years) 5.4 years (3.8 – 7.9 years) 3.8 years (n=148, 2.7 – 6.8 years) 1,253 years 133 PTV/delc, 21 missense, 2 complex indels
Clinical characteristics  Epilepsy   Median age at seizure onset (IQR)   Generalized-onset seizures   Bilateral tonic-clonic seizures   Atonic seizures   Absence seizures   Myoclonic seizures  NDD/autism/intellectual disability  Behavioral features   Sleep disturbances   Anxiety-related disorders   Aggressive behavior	65.4% (n=161/246) -- 66.5% (n=107/161) 46.6% (n=75/161) 32.9% (n=53/161) 18.0% (n=29/161) 0.62% (n=1/161) 78.9% (n=194/246) 75.2% (n=185/246) 37.3% (n=69/185) 11.9% (n=22/185) 5.9% (n=11/185)	82.9% (n=131/158) 2.8 years (2.0 – 3.8 years) 69.5% (n=91/131) 16.8% (n=22/131) 42.0% (n=55/131) 54.2% (n=71/131) 22.9% (n=20/131) 100% (n=138/138) 95.7% (n=132/138) 74.6% (n=103/138) 23.9% (n=33/138) 52.2% (n=72/138)

^aWe included individuals 25 years of age or younger for all cohorts: SYNGAP1-related disorders, Broad epilepsy cohort, Angelman Syndrome, and Rett Syndrome.

^bIndicated by first clinical diagnosis using International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis code for SYNGAP1 (F78A1), as reflected in claims data.

^cProtein truncating variant and deletions (PTV/del) included nonsense variants, frameshift variants, splice site variants, and whole or partial gene deletions.

^dRace and ethnicity information was not available for the insurance claims cohort or the Citizen Health dataset

Comparison of SYNGAP1-RD to other neurodevelopmental disorders

To better characterize the unique phenotypic landscape in SYNGAP1-RD, we leveraged insurance claims data to compare individuals with SYNGAP1-RD to three comparator groups—a broad epilepsy cohort (n=680,076) obtained via G40+ ICD-10 codes, Angelman Syndrome (n=3,697, ICD-10 Q93.51), and Rett Syndrome (n=5,489, ICD-10 F94.2). For comparison, time-stamped ICD-10 codes looking back six years were available for analysis for all insurance claims cohorts.

Longitudinal phenotypic analysis of the clinical landscape

First, we analyzed SYNGAP1-RD through healthcare claims data (n=246), which enabled us to analyze clinical histories including clinical diagnoses, phenotypic features, and treatment patterns. For treatment strategies, we assessed the current medication landscape for anti-seizure medications (ASMs) in addition to medications for behavioral features including anxiety, aggression, and sleep-related disorders.

Next, we assessed clinical histories in a smaller cohort (n=158) focusing on epilepsy trajectories and developmental outcomes from reconstructed medical records. We used a published framework championed by the Epilepsy Learning Health System (ELHS)¹⁹ and Pediatric Epilepsy Learning Health System (PELHS)²⁰ and used previously by our group^21,22 to capture seizure severity on a monthly basis, in which seizure frequencies (SF) are indicated by: multiple daily seizures (>5 per day, SF score = 5), several daily seizures (2–5 per day, SF score = 4), daily seizures (SF score = 3), weekly seizures (SF score = 2), monthly seizures (SF score = 1), and no seizures (SF score = 0). To assess development, we analyzed milestone acquisition. Clinical features and seizure types were captured using the Human Phenotype Ontology (HPO), a framework for phenotypic data harmonization.^23–26 Clinical diagnoses were mapped to the HPO using the Unified Medical Language System (UMLS) crosswalk,²⁷ supplemented by manual curation of seizure-related codes. Cumulative developmental milestone acquisition was plotted in Figure 4 relative the age at which 75% of typically developing children have attained the respective developmental milestone, as reported by the Center for Disease Control and Prevention (CDC).

Figure 4. Developmental skill acquisition in 158 individuals across 1,254 patient-years illustrates the unique features of the SYNGAP1.

(A) The cumulative acquisition of commonly assessed milestones is displayed over the lifespan, for those individuals for which that skill was attained. Vertical dashed lines indicate the age at which 75% of typically developing children have attained the respective developmental milestone, as reported by the Center for Disease Control and Prevention (CDC). The median age at achieving head control was 3 months (n=35, IQR 2 – 6.7 months), rolling over at 5.9 months (n=70, IQR 4.0 – 8.6 months), sit unsupported at 9.8 months (n=73, IQR 8.6 months – 1.1 years), walk with or without assistance at 1.7 years (n=143, IQR 1.5 – 2.3 years), and use short phrases at 4.2 years (n=39, IQR 2.9 – 5.5 years). (B) The proportion of individuals noted to have achieved specific developmental milestones is displayed, with colors representing the different skill domains. The range of developmental milestones achieved spanned from 23% in ability to use verbal communication to more than 95% in ability to communicate using nonverbal language and achieve fine motor abilities.

Comparative ASM effectiveness analysis

Following the retrieval of ASM prescription data and the monthly reconstruction of seizure frequencies, we performed a comparative ASM effectiveness analysis in individuals with SYNGAP1-RD from the Citizen Health dataset or who were seen at CHOP, as described previously.^21,22,28 A total of 94 individuals had ASM prescription and seizure frequency information available.

To determine the comparative effectiveness of ASMs in individuals with SYNGAP1-RD, we analyzed how the SF scores were changing when they were prescribed different medications. For example, if an individual was treated with levetiracetam between two and four years of age, changes in SF scores at that time were compared to SF scores when they were not treated with levetiracetam. Fisher’s exact test was performed to assess how different ASMs are comparatively effective in (1) reducing seizure frequency and (2) maintaining seizure freedom. Only medications that were prescribed to at least 5 individuals were considered in the comparative ASM effectiveness analysis.

RESULTS

Individuals with SYNGAP1-RD can be identified through various healthcare resources

We identified individuals with SYNGAP1-RD through a combination of healthcare claims data, medical record aggregators, and individuals seen at a single tertiary pediatric healthcare center. 246 individuals with SYNGAP1-RD were identified through healthcare claims data, spanning 1,321 cumulative patient-years. Reconstructed medical records of 158 individuals were included from the Citizen Health Natural History Registry (n=138) and Children’s Hospital of Philadelphia (n=20) across a total of 1,253 cumulative patient-years. The median age of inclusion was 6.53 years (IQR 1.5 – 12.5 years) and 5.6 years (IQR 3.8 – 10.0 years) in each respective cohort, and the median observation time was 6 years (IQR 4.8 – 6.3 years) and 5.4 years per individual (IQR 3.8 – 7.9 years) in the claims data and medical record cohorts, respectively. The median age of genetic diagnosis was 3.8 years (n=148 individuals where genetic testing was available, IQR 2.7 – 6.8 years). In the claims dataset, age at diagnosis was not available, but the median age at first SYNGAP1 diagnosis code was 11.2 years (IQR 5.6 – 16.5 years).

The demographic and clinical characteristics across both cohorts were similar, but with a few notable differences (Table 1). For example, 131 of 158 individuals (82.9%) from reconstructed medical records had seizures, compared to 161 of 245 (65.4%) individuals identified through claims data with SYNGAP1-RD. Claims data also had relatively lower documented rates of neurodevelopmental delay, autism, and/or intellectual disability (78.9% compared to 100%) and behavioral features (75.2% compared to 95.7%).

Phenotypic footprints of SYNGAP1 are distinct from other neurodevelopmental disorders

In order to assess disease-specific signatures or phenotypic footprints related to SYNGAP1-RD, we defined a broader epilepsy cohort of 680,076 individuals with a G40+ diagnosis and assessed cumulative diagnoses of G40+ alongside SYNGAP1. When comparing individuals with SYNGAP1 with the broader epilepsy cohort, we found 95 significant associations after correcting for multiple testing (Fig 1A). Of note, individuals with SYNGAP1-RD were more likely to have behavioral abnormalities (Odds ratio (OR) 12.35, 95% CI 9.21 – 16.78), generalized-onset seizures (OR 1.56, 95% CI 1.20 – 2.02), autism (OR 12.23, 95% CI 9.29 – 16.24), and abnormality of higher mental function (including intellectual disability, OR 6.38, 95% CI 4.89 – 8.37). In contrast, individuals with SYNGAP1 were less likely to have motor seizures than those in the broader G40+ cohort (OR 0.273, 95% CI 0.212 – 0.353).

Figure 1. Clinical phenotypes enriched in individuals with SYNGAP1 in comparison to other epilepsies and neurodevelopmental disorders.

Individuals with SYNGAP1-RD (y-axis) were compared to a broader epilepsy cohort of 680,076 individuals with an epilepsy (G40+) diagnosis (A, x-axis), as well as with syndromic comparator groups that phenotypically resembled SYNGAP1: 3,697 individuals with Angelman Syndrome (B, x-axis) and 5,489 individuals with Rett Syndrome (C, x-axis). Features above the diagonal line occur more frequently in individuals with SYNGAP1-RD. Significant associations are shown in red and the dot size represents the -log10(p-value). Compared to the broad epilepsy cohort, behavioral abnormalities (Odds ratio (OR) 12.35, 95% CI 9.21 – 16.78), generalized-onset seizures (OR 1.56, 95% CI 1.20 – 2.02), autism (OR 12.23, 95% CI 9.29 – 16.24), and abnormality of higher mental function (including intellectual disability, OR 6.38, 95% CI 4.89 – 8.37) were enriched in individuals with SYNGAP1. When comparing SYNGAP1 with Angelman and Rett syndromes, behavioral features (AS OR 7.2, 95% CI 5.3 – 9.9, RS OR 5.0, 95% CI 3.7 – 6.8) and autism (AS OR 6.8, 95% CI 5.1 – 9.2; RS OR 4.4, 95% CI 3.3 – 6.8) were more common.

We then contrasted individuals with SYNGAP1-RD with comparator groups of individuals that phenotypically resembled SYNGAP1, particularly during early development, including (1) 3,697 individuals with Angelman Syndrome and (2) 5,489 individuals with Rett Syndrome (Fig 1B and 1C). When comparing SYNGAP1 with Angelman (AS) and Rett (RS) syndromes, behavioral features remained enriched in those with SYNGAP1 (AS OR 7.2, 95% CI 5.3 – 9.9, RS OR 5.0, 95% CI 3.7 – 6.8), as did autism (AS OR 6.8, 95% CI 5.1 – 9.2; RS OR 4.4, 95% CI 3.3 – 6.8). Additional clinical features included higher rates of abnormal verbal communicative behavior (AS OR 3.4, 95% CI 1.8 – 6.2, RS OR 2.4, 95% CI 1.3 – 4.2), increased typical absence seizures (AS OR 2.6, 95% CI 1.7 – 4.0, RS OR 5.3, 95% CI 3.3 – 8.1), and decreased relative frequency of status epilepticus (AS OR 0.68, 95% CI 0.48 – 0.95, RS OR 0.64, 95% CI 0.45 – 0.89).

Longitudinal trajectories demonstrate age-specific phenotypic patterns, including a later seizure onset and ongoing seizures in SYNGAP1

We then assessed clinical features across the age span (Fig 2). We found that the overall clinical presentation of SYNGAP1-RD begins to diverge from comparator groups by the second year of life. When compared to the general epilepsy cohort, we found that individuals with SYNGAP1-RD were more likely to have behavioral abnormalities, first significant between 27 and 30 months (OR 3.00, 95% CI 1.50 – 5.68) and persisting throughout the lifespan. Autistic behavior also became prominent between 27 and 30 months (OR 5.71, 95% CI 2.44 – 11.9). Generalized-onset seizures became significantly enriched (OR 4.05, 95% CI 2.02 – 7.59) after 3 years of age.

Figure 2. Age-related clinical features in SYNGAP1 compared to longitudinal histories of other epilepsies and neurodevelopmental disorders.

The overall clinical presentation of SYNGAP1-RD begins to diverge from comparator groups by the second year of life. (A) The significance of enrichment of key features compared to comparator groups (G40+ in orange, Angelman Syndrome in grey, Rett Syndrome in blue) is plotted across the lifespan. (B) The frequencies of selected features across the lifespan are shown for individuals with SYNGAP1-RD and comparator groups. When compared to the general epilepsy cohort, we found that individuals with SYNGAP1 were more likely to have behavioral abnormalities, first significant between 27 and 30 months (OR 3.00, 95% CI 1.50 – 5.68) and persisting throughout the lifespan. Autistic behavior also became prominent between 27 and 30 months (OR 5.71, 95% CI 2.44 – 11.9). Generalized-onset seizures became significantly enriched (OR 4.05, 95% CI 2.02 – 7.59) after 3 years of age.

The majority of individuals with SYNGAP1-RD have generalized epilepsy, including generalized-onset seizures in up to 70% (frequency range 66.5%, n=107/161 claims data to 69.5%, n=91/131 medical records). The frequency of bilateral tonic-clonic seizures ranged from 16.8% (n=22/131 in medical records) to 46.6% (n=75/161 in claims data), absence seizures in 18.0% (n=29/161) to 54.2% (n=71/131), atonic seizures in 32.9% (n=53/161) to 42.0% (n=55/131), and generalized myoclonic-atonic seizures in up to 43.5% (n=70/161) with myoclonic seizures captured in n=30/131 (22.9%). The median age of seizure onset was 2.8 years (IQR 2.0 – 3.8 years): 6 individuals had infantile onset, 26 in the second year of life, and 40 in the third year of life, and 59 after the third year of life (Fig 3A).

Figure 3. Seizure landscape in SYNGAP1-RD differs from other common genetic epilepsies.

(A) Breakdown of seizure types and cumulative onset across the EMR cohort. Generalized-onset seizures occur in up to 70% (frequency range 66.5%, n=107/161 claims data to 69.5%, n=91/131 medical records). The frequency of bilateral tonic-clonic seizures ranged between 16.8% (n=22/131 in medical records) to 46.6% (n=75/161 in claims data), absence seizures in 18.0% (n=29/161) to 54.2% (n=71/131), atonic seizures in 32.9% (n=53/161) to 42.0% (n=55/131) and generalized myoclonic-atonic seizures in up to 43.5% (n=70/161). (B) Reconstructed seizure frequencies in monthly time bins illustrate the higher frequency and later onset (median 2.8 years, IQR 2.0–3.8 years) of seizures compared to other common genetic epilepsies. Individual monthly seizure frequencies are coded by color (darker = higher seizure frequency); yellow transitions between bars indicate worsening month-to-month seizure frequencies, while blue connectors indicate improving seizure frequencies.

Detailed monthly seizure histories were reconstructed in 13 individuals with SYNGAP1-RD across 154 cumulative patient-years. Epilepsy trajectories of SYNGAP1 are distinct from other genetic epilepsies including STXBP1-related disorders and SCN2A/8A-related disorders (Fig 3B). Of individuals with epilepsy, seizures tended to start after the second year of life and were more likely to be refractory with frequent generalized seizures including absence seizures and atonic seizures. When assessing the median seizure frequency, we found that individuals with SYNGAP1-RD often had seizure frequencies falling within the highest ELHS/PELHS category (>5 per day) due to the prevalence of absence and myoclonic seizures, which tend to occur at higher frequencies than other seizure types.

Developmental outcomes and behavioral features are distinct in SYNGAP1-RD

We mapped milestone acquisition in 158 individuals across 1,254 patient-years. The range of individuals achieving specific developmental milestones spanned from less than 23% for the ability to use verbal communication to more than 95% for the ability to communicate using nonverbal language and achieve fine motor abilities (Fig 4B). For those individuals achieving each specific milestone, the median age at head control was 3 months (n=35, IQR 2 – 6.7 months), rolling over at 5.9 months (n=70, IQR 4.0 – 8.6 months), sitting unsupported at 9.8 months (n=73, IQR 8.6 months – 1.1 years), walking with or without assistance at 1.7 years (n=143, IQR 1.5 – 2.3 years), and using short phrases at 4.2 years (n=39, IQR 2.9 – 5.5 years, Fig 4A).

Almost all individuals with SYNGAP1-RD have behavioral features (95.7%, n=132/138 in medical record cohort), including sleep disturbance in up to 74.6% (n=103/138), anxiety-related behavior in up to 23.9% (n=33/138), and aggressive behavior in up to 52.2% (n=72/138).

The medication landscape in SYNGAP1 demonstrates etiology-specific treatment strategies

Lastly, we assessed the current treatment landscape of SYNGAP1-RD. Leveraging claims data for ASM and behavioral medication prescriptions, we were able to reconstruct the medication histories of 246 individuals across 45 unique medications. For epilepsy therapy (Fig 5A), we found that valproic acid (OR 2.26, 95% CI 1.29 – 3.70) and clobazam (OR 2.58, 95% CI 1.55 – 4.09) were more frequently used in individuals with SYNGAP1 in comparison to the broader cohort of individuals with epilepsy (G40+). The medication landscape for behavioral features in SYNGAP1-RD (Fig 5B) demonstrated that risperidone (OR 5.43, 95% CI 3.47 – 8.18), aripiprazole (OR 3.52, 95% CI 2.05 – 5.69) and guanfacine (OR 2.97, 95% CI 1.76 – 4.75) were more commonly used for managing behavioral symptoms.

Figure 5. The ASM prescription data reveals SYNGAP1-specific treatment approaches.

Medication prescription landscape shows the proportion of individuals prescribed common anti-seizure (A) and behavioral (B) medications across the lifespan. Individuals with SYNGAP1 are more commonly treated with valproic acid (OR 2.26, 95% CI 1.29 – 3.70) and clobazam (OR 2.58, 95% CI 1.55 – 4.09) for epilepsy and risperidone (OR 5.43, 95% CI 3.47 – 8.18), aripiprazole (OR 3.52, 95% CI 2.05 – 5.69) and guanfacine (OR 2.97, 95% CI 1.76 – 4.75) for behavior, which contrasted with treatment strategies frequently used in the broader epilepsy population. (C) Comparative ASM effectiveness analysis (green – significant positive association [P < 0.05 with OR > 1], grey – non-significant association, red – significant negative association [P < 0.05 with OR < 1]). The use of valproate and lamotrigine demonstrated effectiveness in reducing seizure frequencies or maintaining seizure freedom.

In terms of the ASM effectiveness (Fig 5C), we found that valproate (n = 43, P = 0.016, OR 1.33, CI 1.05–1.67) had a significant association with reductions in seizure frequencies as well as with maintaining seizure freedom (n = 43, P = 0.013, OR 1.38, CI 1.06–1.77). Lamotrigine (n = 35, P = 2.91 × 10⁻⁸, OR 1.93, CI 1.53–2.42) also showed significant associations with maintaining seizure freedom. In contrast, levetiracetam (n = 44, P = 2.23 × 10⁻⁶, OR 0.36, CI 0.2–10.58) and clobazam (n = 47, P = 6.43 × 10⁻³, OR 0.67, CI 0.49–0.90) were associated with the lack of seizure freedom.

DISCUSSION

Disease-causing variants in SYNGAP1 result in a recognizable neurodevelopmental disorder characterized by generalized epilepsy, developmental delay, intellectual disability, and behavioral features. However, there is considerable variability in disease course, severity, and functional outcome. As novel precision medicine therapeutics are under development for rare genetic epilepsies, including SYNGAP1, complete characterization of the phenotypic landscape and developmental outcomes of the condition is critical for clinical trial readiness.^{13,14,29–31}

While well-designed natural history studies with validated outcome measures are the ideal precursor to clinical trial design, such studies are expensive, time-consuming, and have difficulties achieving enrollment targets, especially in rare disease.³² The current study lays the groundwork for future natural history studies by leveraging real-world data across multiple healthcare resources to delineate the longitudinal epilepsy and phenotypic landscape of SYNGAP1-RD. The use of EMR and claims data also allows for the comparison of longitudinal phenotypes in SYNGAP1-RD to other similar syndromic comparator groups, which allows us to identify early phenotypic fingerprints of these disorders.

We capitalized on three unique real-world datasets to delineate both the overall burden of clinical features as well as longitudinal symptom trajectories. Our analyses show a significant increase in autism and behavioral abnormalities starting at 27 months, representing the earliest distinguishing features of this disorder, when compared to a general epilepsy cohort. Generalized seizures were also a key feature, but only became significantly increased after 3 years of age. Furthermore, when analyzed relative to two similar syndromic comparator groups—Angelman and Rett Syndromes—the relative increased frequencies of both autism and behavioral features in SYNGAP1-RD held true.

Seizure histories in SYNGAP1-RD are unique among the genetic epilepsies, with relatively later onset (median 2.8 years) and a tendency towards persistent frequent seizures throughout childhood. Most common seizure types were generalized, including absence, atonic and myoclonic. Of note, bilateral tonic-clonic seizures were present in 47% of those with seizures reported in claims data, while only in 17% of those in the EMR datasets. Two possible explanations are that (1) more severe seizures are more likely to be coded in claims data, thus inflating the proportion of significant motor seizures while underrepresenting milder seizure types such as absence, and (2) that non-neurologists often document all seizures with motor signs as “generalized tonic-clonic seizures (GTCs),” introducing a systemic error to the claims dataset.

In our study, we identified an overall lower rate of epilepsy compared to previously published cohort studies.^4,7 This is likely due to two factors—the continual improvement in access to genetic testing has led to earlier and broader testing, including those individuals without seizures and milder phenotypes, and the natural limitations of claims and EMR-based datasets. Historically, only the most severely affected individuals were offered genetic testing, but as it becomes more ubiquitous, we expect to see a widening of the phenotypic spectrum, encompassing more mild cases over time.

Development in SYNGAP1-RD follows a distinct trajectory, with global delays in milestone achievement, however, language acquisition is reliably the most severely affected. In our clinical experience, behavioral concerns are often the most significant issue raised by caregivers of individuals with SYNGAP1-RD, which has previously been demonstrated by smaller case-series³³ and recapitulated here in our large, retrospective, real-world datasets. Behavioral medications being so frequently prescribed also underscores the impact of the behavioral symptoms in this disorder.

While targeted therapeutics are currently under development,^13–15 medical treatment of SYNGAP1-related epilepsy and behavior follows typical treatment practices for other generalized epilepsies and behavioral disorders. Seizures are most commonly treated with clobazam and valproic acid. Behavioral medications are very frequently prescribed, with the most common choices being risperidone, guanfacine, and aripiprazole. Comparative effectiveness analysis confirmed the efficacy of valproic acid and lamotrigine for reducing seizure frequency and maintaining seizure freedom. This is consistent with our experience in clinical practice and is intuitive as these are both effective medications for generalized seizures. However, neither clobazam nor levetiracetam were shown to be effective in this cohort. This may be related to a lack of power to detect a significant reduction in seizures, but may also be due to the clinical characteristics of these individuals. In particular, levetiracetam is often prescribed for eyelid myoclonia, so our results may be reflective of the inherent high frequency and treatment-resistance of this seizure type. Additionally, one unique feature about the claims dataset is the relatively low capture of cannabidiol prescriptions; this appears to be specific to the data vendor used in these analyses and may be due to specialty pharmacy blocking. While the current treatment strategies are reactionary—treating symptoms as the arise—identifying individuals with SYNGAP1-RD early will be critical once clinical trials for targeted therapies are available.

While the primary mechanism of disease in SYNGAP1-RD is haploinsufficiency, prior studies have suggested possible genotype-phenotype correlations, including the lower prevalence of epilepsy in individuals with variants in the SH3-binding motif,¹² the relatively milder presentations in individuals with loss of function variants in exons 1–4 compared to exons 5–19,^7,12 and the relative pharmacosensitivity of the epilepsy in individuals with variants in exons 4 and 5.⁶ The variant spectrum from medical record data included 134 protein-truncating variants, 21 missense variants, and two complex or in-frame indels. Given the limitations of claims data, we were unable to link the phenotypic information to a genetic testing report. However, techniques for tokenization of the data are under development which will make linking claims data to genetic data feasible, greatly enhancing the utility of these large real-world datasets.³⁴ This would also allow for greater harmonization across diverse data sets, ensuring individuals contained in each dataset could be identified and linked, enhancing the quality of their representation.

Due to relatively large sample sizes, claims data can provide useful insights into longitudinal disease trajectories; however, the data is sparse and subject to unique challenges.^35,36 These datasets are prone to errors in medical coding, and any errors are often perpetuated due to the tendency to “copy forward” diagnosis codes from prior encounters. Furthermore, while we are relatively confident that clinical features coded at high frequencies are accurate, the absence of coding of a feature or symptom cannot be taken to mean that feature was not present, as not all clinical symptoms are captured in encounter diagnosis codes. This likely accounts for the differences in frequencies of given clinical features between the claims dataset and the combined electronic medical record dataset—for example, epilepsy was reported in 82.9% of cases obtained through medical records, but only 65.4% of those obtained through insurance claims data.

Taken together, we surveyed the clinical landscape of SYNGAP1-RD and contrasted the phenotypic signature and current state of treatment strategies with other epilepsies and neurodevelopmental disorders, including Angelman Syndrome and Rett Syndrome. This analysis underscored the relatively later onset of a generalized epilepsy with very prominent behavioral features, autism, and significant impairments in expressive language. The integration and analysis of real-world data from various healthcare databases enabled us to leverage varying scopes of data granularity and complexity, demonstrating the utility of large-scale, computational frameworks to understand the prevalence of clinical features and outline the longitudinal landscape of rare diseases. Delineation of disease-specific trajectories in SYNGAP1-RD will remain critical for prospective natural history studies, clinical trial-readiness, and future precision medicine advances.

DATA AVAILABILITY

De-identified aggregate data will be available upon written request to the corresponding author. The raw data that support the analyses presented here contain sensitive and protected health information for participants and are therefore not openly available.