Unsupervised clustering of a deeply phenotyped cohort of adults with idiopathic generalized epilepsy

Idiopathic generalized epilepsy (IGE) in adults comprise juvenile myoclonic epilepsy (JME), juvenile absence epilepsy (JAE), and epilepsy with generalized tonic-clonic seizures alone (EGTCS), which are defined by their seizure types but also cover a broad endophenotype of symptoms. Controversy exists on whether adult IGE is a group of distinct diseases or a clinical spectrum of one disease. Here, we used a deeply phenotyped cohort to test the hypothesis that IGE comprises three distinct clinical entities.

Patients (>18 years old) with IGE were recruited between 2016 and 2020 at Odense University Hospital and the Danish Epilepsy Center. Complete data were available for basic demographics, imaging, social status, and treatment response. Subjects were offered neuropsychological screening (including symptoms of psychiatric disease). Electroencephalograms (EEGs) were reanalyzed, and missing data were imputed. After selecting the features and normalizing the data, the dataset and an identical randomized dataset were subjected to k-means cluster analysis.

The dataset comprised 502 patients and 22 distinct nonoverlapping clinical features. Elbow and gap analyses revealed an optimal number of clusters of 1; the features with the highest eigenvalues were age at diagnosis and self-reported executive dysfunction. Applying k-means clustering yielded three low-quality clusters as assessed by silhouette score (mean = .400 ± .01). The corresponding mean silhouette score for the randomized control dataset was .390 (±.002). Age at diagnosis was associated with the epileptic discharges on EEG and treatment response. Self-reported executive dysfunction showed associations with psychiatric symptoms and impulsivity. JME, JAE, and EGTCS were loosely associated with the clusters (k = .088, p = .002) but showed a specific distribution within a matrix defined by age at diagnosis and self-reported executive dysfunction.

IGE in adults is best described as a continuum of symptoms, where age at diagnosis and executive dysfunction are two main factors explaining most of its clinical variability. The seizure-defined syndromes cover different patient groups within the clinical spectrum.