Benchmarking long-read sequencing approaches to resolve facioscapulohumeral dystrophy locus complexity

Facioscapulohumeral dystrophy (FSHD) is primarily associated with contraction of the D4Z4 macrosatellite array at the 4q35 locus. Unaffected individuals carry 11-150 D4Z4 repeats, whereas ∼95% of FSHD patients (FSHD1) exhibit a contraction to 1-10 units, along with reduced DNA methylation. In another ∼3% of patients (FSHD2), the disease results from a digenic mechanism associated with the presence of a pathogenic variant in the SMCHD1 gene, leading to the epigenetic deregulation of the 4q35 locus. However, 1%-2% of clinically diagnosed patients lack a defined genetic cause, highlighting diagnostic gaps. In prior work, we identified >70 patients, clinically diagnosed with FSHD and carrying a complex structural variant of the 4q35 or 10q26 loci. A potential pathogenicity of these structural variants was evoked in some cases, in the absence of other FSHD-associated genetic features. Given their diagnostic relevance, here we performed detailed structural analyses of these rearrangements in seven representative cases carrying different structural variants of the 4q35 or 10q26 loci using high-resolution long-read sequencing technologies (Oxford Nanopore and PacBio) and suspected of FSHD. By comparing the advantages and limitations of several methodological long-read sequencing strategies, we resolved the architecture and methylation patterns across the 4q35 and 10q26 loci at the nucleotide level. We show that duplicated alleles arise from intrachromosomal recombination between LSau elements contained within D4Z4 and distal subtelomeric β-satellite elements, producing variable deletions within the proximal D4Z4 region, with breakpoints differing among patients. These complex structural variants are not detectable using standard technologies, such as Bionano Optical Genome Mapping, and require manual curation for identification during routine molecular diagnosis procedures. Importantly, determining the pathogenic relevance of these rearrangements necessitates integration of structural and epigenetic features typically associated with FSHD. Our results underscore the importance of in-depth molecular characterization for patients with clinical FSHD who test negative for FSHD1/FSHD2 by conventional diagnostic methods. We also show that structural variants might be considered as likely to be pathogenic, in the absence of an SMCHD1 variant. Overall, as structural variants at 4q35 are increasingly identified in patients clinically diagnosed with FSHD, their comprehensive analysis is crucial to refine diagnosis, guide genetic counselling and, ultimately, improve clinical care for individuals clinically suspected of FSHD but presenting with an atypical molecular profile.