To our knowledge, this is the first study to compare the prevalence of core clinical features of a single rare genetic disorder, in this case SLC6A1, found in databases generated by medical professionals and content experts to data in a caregiver-reported patient registry, Simons Searchlight. With any retrospective data review, investigators are limited by the structure of existing datasets. As demonstrated in our large retrospective provider-reported dataset, detailed clinical phenotyping data are limited [4]. In that cohort, clinical data were available on the majority of the group for conditions including epilepsy, developmental delay, and autism spectrum disorder; however, less than half had data available on common comorbid conditions such as ADHD, hypotonia, and movement disorders. In addition to missing data elements, the depth of data was also limited. Qualitative reports of broad diagnoses such as developmental delay are provided rather than detailed standardized testing information or categorization of severity of deficits. To simplify analysis of the retrospective data, we combined developmental delay and intellectual disability. This is a major limitation of retrospective data collection because not all who have developmental delay will also meet criteria for intellectual disability and intellectual disability may be a confounding variable. Standardized validated measures like the Vineland Adaptive Behavior Scales would allow for more precise quantitative data to be collected from individuals, which may highlight a paradigm shift in the traditional in-person evaluations for rare disorders to utilizing a combination of in person and virtual assessments [11]. In addition, medical evaluations may consider including standardized measures such as the Vineland to collect valuable data that also can be easily translated across various clinical sites.

Moreover, it becomes difficult to assess whether there are overlapping subjects being reported in larger case series by different investigators without the incorporation of a common identifier such as the global unique identifier (GUID). Utilization of the GUID across more studies in rare disease, as is done in Simons Searchlight, could allow for matching subjects between studies, which would increase the portability and potentially increase the power of clinical datasets in rare disease populations. Given the rarity of SRD, it is plausible that a single individual is included in more than one database, and with insufficient phenotypic and demographic data, it is difficult to rule out the presence of overlapping patients within these datasets, especially if there are recurring genotypes. If individuals are duplicated within the combined provider-reported dataset, there may be an inappropriate weighting of their associated clinical features, skewing the data. In addition, these datasets are often retrospective cohort studies, and there is a lack of prospective or longitudinal data collection.

We sought to determine if caregiver report data from a relatively large registry, Simons Searchlight, would show consistency with what has previously been reported in the literature based on medical provider data. Simons Searchlight verifies genetic testing results to confirm pathogenicity of the variants, collects medical records, and also assigns unique identifiers to avoid duplication of subjects. Importantly, they collect data prospectively using validated questionnaires that are commonly used in clinical practice and other research studies and provide more behavioral phenotypic information. Though the caregiver-reported dataset (Simons Searchlight) is a smaller sample, data collection is more rigorous and standardized, and there are fewer missing data points. In the case of the SLC6A1 cohort of Simons Searchlight, the sample size is still small and only a single time-point is available. As additional patients are identified and multiple time-points are collected longitudinally, the power of this registry will increase. Even still, neither of these observational study designs are sufficient to supplant a prospective natural history study, but rather, provide insights on the spectrum of a rare disease and inform the design of prospective studies by collecting an unbiased, structured dataset on all individuals.

The concordance of the prevalence of epilepsy and autism spectrum disorder between the two groups supports that these are two of the core clinical features of SRD. Further, there were the fewest number of missing data points for these two clinical features in the provider-reported dataset. Findings on the seizure survey and on standardized ASD instruments within the Simons Searchlight patient registry support the presence of epilepsy and of autistic traits such as repetitive behaviors, though clinical diagnoses of ASD were discordant with scores on ASD instruments. There was no difference in scores between those with ASD and those without ASD. However, questionnaire screeners are notoriously limited in their ability to distinguish ASD in the presence of intellectual disability and other comorbidities. Further clinician-generated evaluations, such as those including the Autism Diagnostic Observation Scale in individuals of sufficient mental age, are needed to evaluate whether individuals with SRD have ASD or if these ASD screening tools are showing elevated scores for autistic traits, such as repetitive behaviors. The relatively higher reports of DD/ID per caregiver report are also reflected in both early and late developmental delays according to developmental milestones and results of the standardized Vineland interview.

In contrast, there was higher prevalence of hypotonia and movement disorders in the caregiver-reported dataset, with hypotonia reaching statistical significance. The high rate of missing data points within these clinical features in the provider-reported data highlights a limitation of the provider-reported dataset, in which the phenotypic descriptions are not standardized or clinically confirmed. Further, published case reports often highlight certain features of a disorder while omitting others that are relevant to the audience of the journal in which the publication appears. A few potential explanations include (1) Simons Searchlight, because of the smaller sample, is subject to sampling bias; (2) these features were present in the provider-reported data but not included in the publication or database; or (3) these features were omitted from the provider-reported data because they were not present in individuals. Because SRD was initially described in epilepsy journals, many of the early descriptions of the disorder included deep phenotyping of the seizure semiology and developmental abilities, which are a significant source of morbidity in developmental epileptic encephalopathies. While features such as hypotonia are quite common among neurodevelopmental disorders, the reports in the literature may omit these details. As the phenotypic spectrum of genetic developmental epileptic encephalopathies expands, additional features, such as movement disorders like tremor or stereotypy, are emerging as a relatively common feature of these disorders. Patient registries with standardized assessment tools applied to the entire sample have the potential to uncover new features of disorders or identify clinical outcome measures that could inform a future clinical trial. Finally, harmonization of tools across patient registries and agreement upon a set of common data elements and unique common identifiers would facilitate the compilation of multiple datasets, improve portability of clinical data, and potentially expedite important discoveries that could make rare disease clinical research more efficient.