ABSTRACT
Rationale: Severe COVID-19 pneumonia can be complicated by secondary bacterial or fungal infections, but their clinical distinction from isolated SARS-CoV-2 infection is challenging, especially with the more restricted practices regarding invasive diagnostics in patients with COVID-19. We sought to comprehensively screen for secondary infections by DNA pathogens (bacterial, fungal or viral) with a non-invasive, culture-independent metagenomic approach (microbial cell-free DNA sequencing-mcfDNA-Seq), and also examine for the biologic impact of circulating mcfDNA on the host response in COVID-19. Methods: We prospectively enrolled 42 hospitalized patients with COVID-19 and compared them with a historical cohort of mechanically-ventilated patients with culture-positive (n=27) vs. culture-negative pneumonia (n=40) or no clinical infection (n=16 controls). From plasma samples, we performed mcfDNA-Seq with the Karius test (Karius, Inc) and measured 10 host-response biomarkers of innate immunity and epithelial/endothelial injury (IL-6, IL-8, IL-10, RAGE, TNFR1, Angiopoietin-2, Procalcitonin, Fractalkine, Pentraxin-3, ST2). We compared mcfDNA-Seq between clinical groups and examined associations of mcfDNA and biomarker levels with linear regression models. Results: McfDNA-Seq was successful in 33/42 (79%) baseline samples from patients with COVID-19, with nine samples failing QC requirements. McfDNA was detectable in 21/33 (64%) of COVID-19 samples, a proportion significantly lower to culture-positive pneumonia (96%), higher than uninfected controls (31%) and similar to culture-negative pneumonia (56%) (between-groups Fisher's exact p<0.001). A similar distribution was seen for mcfDNA levels, with mcfDNA load in COVID-19 being similarly distributed as non-COVID culture-negative pneumonia (Figure 1A). Among patients with COVID-19, mcfDNA was significantly associated with higher levels of host-response biomarkers (Figure 1B), with stronger effect sizes observed for biomarkers of innate immunity (IL-8 and ST2) and bacterial infections (procalcitonin and pentraxin-3). Conclusions: Plasma metagenomics in patients with COVID-19 revealed mcfDNA load of similar magnitude as in critically-ill patients without COVID-19 with clinically suspected infection but negative microbiologic cultures. The significant associations of mcfDNA with host inflammation support the biological relevance of detectable circulating mcfDNA. Our preliminary results warrant further study of secondary infections in hospitalized patients with COVID-19 to define the clinical utility of noninvasive molecular diagnostics for antimicrobial treatment guidance.