Long-read genome sequencing resolves complex structural variants in exome-negative developmental and epileptic encephalopathy

Background and Objectives: Developmental and Epileptic Encephalopathies (DEEs) encompass a clinically and molecularly heterogeneous group of mainly genetic disorders. Despite standard genetic testing, approximately two-thirds of patients remain undiagnosed. Pathogenic noncoding and structural variants (SVs), particularly complex rearrangements that are difficult to detect or resolve using short-read sequencing, may contribute to this diagnostic gap. To evaluate the diagnostic contribution of long-read genome sequencing (lrGS), we recruited a cohort of 124 families with sporadic DEE who remained unsolved by prior ES. Methods: Trio lrGS (Oxford Nanopore or PacBio HiFi) was performed, complemented zby short-read genome sequencing (srGS) and mRNA sequencing (blood or fibroblast) of index cases, where available. Comprehensive variant analysis included small nucleotide variants (SNVs), indels, SVs, and short tandem repeats. To identify the diagnostic yield of each method, we compared lrGS, srGS and ES re-analysis, and evaluated the additional value of RNA sequencing. Results: We established a diagnosis in 21 of 124 families (16.9%). Fourteen variants were retrospectively detectable in prior ES., Two cases with non-coding variants in SCN1A and RNU2-2 were detectable by both srGS and lrGS. Five SVs also required genome sequencing (5.6% of the total cohort), three of which could be detected, but not confidently resolved by srGS due to breakpoint complexity, repeat-mediated rearrangements, and large copy-neutral events: LrGS enabled accurate characterization of Alu-mediated CDKL5 and STXBP1 deletions and a MEF2C deletion with large inversion. A copy-neutral SCN1A inversion was well characterizable by both srGS and lrGS. One balanced translocation t(X;17) disrupting CDKL5 was resolved by lrGS (srGS unavailable). RNA sequencing confirmed pathogenicity of two SVs including t(X;17), the MEF2C deletion with inversion, and two CNVs. In addition, ZNF569 and ZNF570 emerged as candidate epilepsy genes from our cohort. Discussion: In ES-negative DEE, the incremental diagnostic yield of genome sequencing was modest, while lrGS provided particular value for the structural resolution of complex rearrangements, especially those involving repetitive sequences. The finding that two-thirds of diagnoses were retrospectively detectable in prior ES highlights the substantial value of systematic reanalysis. Further gains from genome sequencing may increasingly depend on improved interpretation of non-coding variation and integration of long-read-specific information. Our findings indicate that complex SVs represent the class for which lrGS currently yields the most distinctive diagnostic information.