1097. Microbial Cell Free DNA Sequencing for Prediction of Culture-Negative Infection Events in Children with Cancer

Abstract Background Culture-independent diagnostics may help diagnose or predict infection; microbial cell free DNA sequencing (mcfDNA-seq), can detect a wide range of pathogens directly from plasma. Immunocompromised children who develop febrile neutropenia (FN) without documented bloodstream infection (BSI) may have undiagnosed bacterial infection, but identification of this is difficult, and the proportion of such episodes is unknown, as is the relative contribution of non-bacterial etiologies. We analyzed mcfDNA-seq results in a convenience sample of FN cases without known etiology. Methods Participants were < 25 years of age and undergoing treatment for cancer. Remnant plasma was prospectively obtained and stored. Samples from Days 0 and -1 underwent mcfDNA-seq by Karius Inc., reported in molecules per microliter (MPM) of plasma. Samples from participants without impending or recent fever or infection were also tested. Results There were 8 episodes in 7 patients; 4 (50%) had a common bacterial pathogen identified by mcfDNA-seq on Day 0 (Table 1). In 2 (50%) of these cases, the same organism was also identified on Day -1, at a lower concentration. One fungal pathogen was identified prior to and at onset of FN. A common bacterial pathogen was identified in 3/64 (5%) control samples from the population. Culture-negative sepsis was the final diagnosis in one episode; Streptococcus mitis, an important cause of neutropenic sepsis, was found in Day 0 and Day -1 samples. In an episode where E. coli was identified by mcfDNA-seq, FN recurred after antibiotic discontinuation. Table 1. Quantitative mcfDNA-seq Results for Prediction & Diagnosis of Febrile Neutropenia Episodes Conclusion In this sample of culture-negative FN episodes in pediatric patients leukemia, mcfDNA-seq identified a bacterial pathogen in 50% of cases. The same organism was identifiable on the day prior to FN in 50% of cases, suggesting that predictive testing might be feasible. Disclosures Asim A. Ahmed, MD, Karius (Employee) Desiree D. Hollemon, MSN, MPH, Karius inc (Employee) Charles Gawad, MD PhD, Karius inc (Grant/Research Support) Joshua Wolf, MBBS, PhD, FRACP, Karius inc (Grant/Research Support)