Clinical features and genotype-phenotype correlations in epilepsy patients with de novo DYNC1H1 variants

DYNC1H1 variants are involved on a disease spectrum from neuromuscular disorders to neurodevelopmental disorders. DYNC1H1-related epilepsy has been reported in small cohorts. We dissect the electroclinical features of 34 patients harboring de novo DYNC1H1 pathogenic variants, identify subphenotypes on the DYNC1H1-related epilepsy spectrum, and compare the genotype-phenotype correlations observed in our cohort with the literature.

Patients harboring de novo DYNC1H1 pathogenic variants were recruited through international collaborations. Clinical data were retrospectively collected. Latent class analysis was performed to identify subphenotypes. Multivariable binary logistic regression analysis was applied to investigate the association with DYNC1H1 protein domains.

DYNC1H1-related epilepsy presented with infantile epileptic spasms syndrome (IESS) in 17 subjects (50%), and in 25% of these individuals the epileptic phenotype evolved into Lennox-Gastaut syndrome (LGS). In 12 patients (35%), focal onset epilepsy was defined. In two patients, the epileptic phenotype consisted of generalized myoclonic epilepsy, with a progressive phenotype in one individual harboring a frameshift variant. In approximately 60% of our cohort, seizures were drug-resistant. Malformations of cortical development were noticed in 79% of our patients, mostly on the lissencephaly-pachygyria spectrum, particularly with posterior predominance in a half of them. Midline and infratentorial abnormalities were additionally reported in 45% and 27% of subjects. We have identified three main classes of subphenotypes on the DYNC1H1-related epilepsy spectrum.

We propose a classification in which pathogenic de novo DYNC1H1 variants feature drug-resistant IESS in half of cases with potential evolution to LGS (Class 1), developmental and epileptic encephalopathy other than IESS and LGS (Class 2), or less severe focal or genetic generalized epilepsy including a progressive phenotype (Class 3). We observed an association between stalk domain variants and Class 1 phenotypes. The variants p.Arg309His and p.Arg1962His were common and associated with Class 1 subphenotype in our cohort. These findings may aid genetic counseling of patients with DYNC1H1-related epilepsy.