D6. Other (e.g., host response biomarkers, molecular imaging, metabolomics/proteomics, etc.)
Late Breaking Abstract Submission
Manfred Grabherr, PhD
Faculty of Medicine
Uppsala University
Uppsala, Uppsala Lan, Sweden
Disclosure: Nothing to disclose
Steve Miller, MD, PhD
Professor and Director
Department of Laboratory Medicine, University of California, San Francisco, San Francisco, CA, USA
San Francisco, CA
Disclosure: Nothing to disclose
Shaun Arevalo, BS
Clinical Laboratory Technician
UCSF
San Francisco, CA
Disclosure: Nothing to disclose
Kelsey Zorn, MHS, BA
Clinical Research Coordinator
UCSF
San Francisco, CA
Disclosure: Nothing to disclose
Kathleen Harriman, PhD, MPH, RN,
Section Chief
California Department of Public Health
San Francisco, CA
Disclosure: Nothing to disclose
Sharon Messenger, PhD
Division Supervisor
California Department of Public Health
Richmond, CA
Disclosure: Nothing to disclose
Samuel Dominguez, MD, PhD
Associate Professor
University of Colorado, School of Medicine
San Francisco, California
Disclosure: Nothing to disclose
Debra Wadford, PhD
Laboratory Director
California Department of Public Health
Richmond, CA
Disclosure: Nothing to disclose
Kevin Messacar, MD
Assistant Professor
University of Colorado
Denver, Colorado
Disclosure: Nothing to disclose
Background: Since 2014 there have been global biennial outbreaks of acute flaccid myelitis (AFM), a rare but severe “polio-like” illness of as yet-unknown etiology primarily affecting children. Enteroviruses (EVs), especially EV-D68 and EV-A71, have been implicated in association with AFM cases, but proving causality has been difficult as EVs are rarely isolated from cerebrospinal fluid. In addition, early identification of EV-associated AFM is challenging given that the diagnosis is reliant on potentially subjective clinical and radiological criteria with no specific biomarkers described to date.
Methods: We leveraged existing and newly generated data from a clinical CSF metagenomic assay for pathogen identification at University of California, San Francisco (UCSF) to interrogate the host response at the transcriptome level by RNA sequencing (RNA-Seq). These transcriptome RNA-Seq data were used to create statistical classification models to discriminate among viral infections that have been linked to AFM, including EV-D68, EV-A71, West Nile virus, and Powassan virus. The dynamic range of CSF cellularity (0 to >106 cells/mL), resulting in varying trancriptome coverage, as well as technical variation across samples required the development and validation of novel normalization techniques. In total, we analyzed ~50 CSF samples split into independent training and test sets.
Results: We were able to demonstrate a distinct signature of AFM that was able to predict the virus associated with AFM in blinded test samples with >80% accuracy. The key transcriptional features that best discriminated EV-A71 from EV-D68-associated AFM involved protein targeting, viral transcription, viral gene expression, and translation initiation pathways.
Conclusion:
Here we demonstrate a novel approach to diagnosis of AFM that relies on host transcriptional biomarkers from cerebrospinal fluid. In the future, this method might allow earlier diagnosis of AFM to drive appropriate therapies and vaccines and predict patient outcomes, as well as guide research studies on the pathophysiology of EV-associated AFM.