Skip to main content

Conference Program

Subpage Hero

Loading

Systematic reanalysis of clinical genome sequencing data in a cohort of acute care patients. 

Laboratory Genetics and Genomics
  • Primary Categories:
    • Clinical- Pediatric
  • Secondary Categories:
    • Clinical- Pediatric
Introduction:
Growing evidence supports the utility of clinical genome sequencing (GS) in the diagnosis of pediatric acute care and outpatient populations with suspicion of a rare genetic disorder. Despite recent technological advancements, many cases remain unresolved and approaches to close this diagnostic gap are needed. In this study we evaluated the impact of systematic reanalysis of GS data in a cohort of newborns admitted to a neonatal intensive care unit in the United States. 

Methods:
Retrospective research analysis of the GS data was performed under a waiver of consent obtained from the Western Copernicus Group Institutional Review Board. Clinical GS data from 351 neonates initially assessed as part of the NICUSeq clinical trial was reanalyzed for this study. The time from original test report to reanalysis ranged from 3-5 years. Reanalysis was initiated from archived FASTQ files that were aligned and assessed for variants using the DRAGEN DNA pipeline version 4.0 on GRCh38. Annotation, tertiary analysis, and variant prioritization were performed using Emedgene version 32. Small variants including single-nucleotide variants (SNVs), and small insertions and deletions, copy-number variants (CNVs), structural variants (SVs), short tandem repeats (STRs), and mitochondrial genome (mtDNA) variants were included in the reanalysis. Human phenotype ontology terms describing the infants’ clinical presentation were obtained by reviewing the medical notes summary on the original clinical GS test reports. Genomic analysts performed a streamlined review of variants identified by the Emedgene explainable AI (xAI) model and variants predicted to be potentially damaging using a set of predefined filters. The number of prioritized variants tagged as ‘Most Likely’ by the xAI model per proband ranged from 4-23, with a mean of 12. Any new reportable variants were evaluated based on ACMG guidelines, and if warranted an amended report was issued to ordering clinicians following standard clinical laboratory procedures. 

Results:
First, previous clinical cases with diagnostic findings (n=91) were run through the reanalysis pipeline to assess its performance. All pathogenic/likely pathogenic (P/LP) variants were recovered, and 73 of 78 small variants (93.6%), and 32 of 34 copy-number variants (94.1%), were prioritized by the Emedgene xAI v32 as ‘Most Likely’ to be disease associated. 

New diagnostic findings were identified in 14 of 351 cases (~4%). Seven of 67 reports with prior variants of uncertain significance (VUS) (10.5%) and seven of 193 previously negative reports (3.6%) were upgraded to Positive, i.e. they were updated to include a diagnostic P/LP variant. The majority of report updates (60%) were due to the availability of new peer-reviewed literature supporting the gene-disease or variant-disease associations. The remaining report updates were enabled by improved variant calling (6.7%), updated variant annotations from ClinVar (20%), new clinical information (6.7%), and additional expert review (6.7%). In one case, improved variant calling by the reanalysis pipeline identified a 4.2 kb deletion of the mitochondrial genome, highlighting the value of reanalyzing GS data starting from FASTQ files as opposed to only using updated annotation and literature sources.  

In an additional six cases, VUS variants in genes with emerging clinical evidence were identified. These VUS variants will be monitored over time as new information for the gene, variant, and/or individual’s clinical presentation may become available.

Conclusion:
Systematic reanalysis of a cohort of early-onset rare disease case genomes resulted in a 4% increase in diagnostic yield. This study provides proof-of-concept that systematic reanalysis of historical acute care cases is feasible, has high impact and requires minimal manual review by genomic scientists. Future efforts to streamline deployment and scalability may enable widespread use of these approaches by other laboratories and provide additional diagnoses to individuals with rare diseases, thus ending their diagnostic odyssey. 

Agenda

Sponsors