Skip to main content

Conference Program

Subpage Hero

Loading

Automated reanalysis of clinical genomic data in undiagnosed patients

Laboratory Genetics and Genomics
  • Primary Categories:
    • Genomic Medicine
  • Secondary Categories:
    • Genomic Medicine
Introduction:


Genomic reanalysis has been shown to identify newly reported variants that are causative for disease. Patient phenotypes can also change over time, particularly in children, making it beneficial to periodically reanalyze negative cases to capture new diagnostic insights. However, clinical genomic data for undiagnosed patients are rarely routinely reanalyzed.  Barriers include data silos that restrict clinicians’ access to raw genomic data, high costs for clinical labs that maintain the data, and the lack of obligation or expectation for labs to provide reanalysis services.

The Children’s Rare Disease Collaborative (CRDC) at Boston Children’s Hospital has integrated raw genomic data and structured annotated variants from over 15,500 patients and their family members, along with human phenotype ontology terms manually entered by clinicians/researchers and automatically pulled and updated from the electronic health record. These data are accessible on a clinician-facing analysis platform, allowing for detailed case-by-case review. This infrastructure empowers systematic re-analysis of all undiagnosed patients across all providers within the institution.

Methods:
Leveraging the CRDC genomic analysis infrastructure, we developed a semi-automated pipeline that enables efficient reanalysis of clinical sequencing data. This pipeline requires minimal manual intervention, allowing us to systematically identify novel variants of interest and return them to clinicians for further review. The pipeline leveraged and integrated multiple phenotype-genotype association tools as well as custom libraries to generate a shortlist of variants, which are further filtered down to candidate variants by a central review board. Those candidate variants were subsequently returned to patients’ providers, who arrange for clinical confirmation before returning results to patients and integrating the result into the electronic health record.

Results:
We performed systematic reanalysis for five years’ worth of clinical sequencing data and successfully uncovered novel diagnostic variants. We identified practical challenges including ordering providers who left the institution, international patients who were difficult to recontact, and patients who had died. We will report on the overall diagnostic yield and impact of these results on patient care, the resources required to support reanalysis, and the potential opportunities to apply our experience to other settings.

Conclusion:
Clinical reanalysis is essential in providing continuous improvements in diagnosis and patient care.  The development of an automated or semi-automated pipeline for re-analyzing patient genomic data is foundational, when combined with access to updated clinical phenotypes, to enable genomic reanalysis at scale.

Agenda

Sponsors