Long-read RNA sequencing in cardiomyopathies: A new approach for genetic diagnostic with strong potential?

Introduction

Dilated cardiomyopathies (DCM) are inherited diseases, for which genetic diagnostic conditions patient follow-up and family care (major gene TTN). However, current DNA Short Read (SR DNA-Seq) based sequencing diagnostic yield is only 15–25% suggesting methodological limitations. This lack might be mitigated by a RNA Long-Read (LR-RNA-Seq) approach since it allows to detect elusive DNA rearrangements and cryptic aberrant splicing causative mutations, representing an attractive strategy to improve diagnostic yield.

Objective

To evaluate feasibility of whole transcriptome LR-RNA-Seq for detection of Single Nucleotide Variants (SNVs) in coding exons and short/complex DNA Structural Variants (SV).

Method

We included 26 DCM cases with cardiac RNA available (transplanted hearts). LR-RNA-Seq was conducted on a PromethION24 (ONT) after cDNA library preparation. SNVs and SV were called using a LR dedicated bioinformatic tool (Clair3). True variants calling thresholds were inferred from replication using DNA Sanger sequencing of the same samples.

Results

At total, samples were covered with 11.6M reads (± 6.9 M) among which 85% were long (> 400 pb) and thus used for alignments. We align 90% of this reads, whose mean size was 1094 pb (± 80). Among 33 putative LR-RNA-Seq variants selected within an extended range of Clair3 quality score, 11 were confirmed in DNA, allowing to defined 3 key parameters witnessing for high prior probability variants: total depth > 15X; minor allele frequency > 0.2 and alternative allele depth > 5X (100% sensitivity; 86 % specificity for variant identification). All 3 causal SNV already diagnosed were detected (100%). We also detected 2 new causative truncating variants in the TTN gene in 2 other patients. Finally, we identified a patient with rare monoallelic variants overlapping with TTN locus, suggesting hemizygosity or consanguinity. Genomic DNA qPCR demonstrated haploidy compared to a control diploid locus, strongly suggesting DCM causative TTN deletion in this patient. Overall, the diagnostic yield of LR-RNA-Seq was 23,0 % (6/26), at the upper limit of the reported yields for DCM.

Conclusion

Long-read RNA seq appear to be an efficient method to detect coding variants in expressed genes, as well as structural variations. Ongoing analysis of splicing and NMD-related variants in our LR-RNA-Seq data might further increase diagnostic yield. LR-RNA-Seq could become a key strategy to complement, or even overwhelm, SR DNA-Seq in genetic diagnosis.