Noninvasive diagnosis of secondary infections in COVID-19 by sequencing of plasma microbial cell-free DNA

Abstract

Background: Secondary infection (SI) diagnosis in COVID-19 is challenging, due to overlapping clinical presentations, practical limitations in obtaining samples from the lower respiratory tract (LRT), and low sensitivity of microbiologic cultures. Research Question: Can metagenomic sequencing of plasma microbial cell-free DNA (mcfDNA-Seq) help diagnose SIs complicating COVID-19? Study Design and Methods: We enrolled 42 inpatients with COVID-19 classified as microbiologically-confirmed SI (Micro-SI, n=8), clinically-diagnosed SI (Clinical-SI, n=13, i.e. empiric antimicrobials), or no clinical suspicion for SI (No-Suspected-SI, n=21) at time of enrollment. From baseline and follow-up plasma samples (days 5 and 10 post-enrollment), we quantified mcfDNA for all detected microbes by mcfDNA sequencing and measured nine host-response biomarkers. From LRT samples among intubated subjects, we quantified bacterial burden with 16S rRNA gene quantitative PCR. Results: We performed mcfDNA-Seq in 82 plasma samples. Sequencing was successful in 60/82 (73.2%) samples, which had significantly lower levels of human cfDNA than failed samples (p<0.0001). McfDNA detection was significantly higher in Micro-SI (15/16 [94%]) compared to Clinical-SI samples (8/14 [57%], p=0.03), and unexpectedly common in No-Suspected-SI samples (25/30 [83%]), similar to detection rate in Micro-SI. We detected culture-concordant mcfDNA species in 13/16 Micro-SI samples (81%) and mcfDNA levels tracked with SI outcome (resolution or persistence) under antibiotic therapy. McfDNA levels correlated significantly with LRT bacterial burden (r=0.74, p=0.02) as well as plasma biomarkers of host response (white blood cell count, IL-6, IL-8, and SPD, all p<0.05). Baseline mcfDNA levels were predictive of worse 90-day survival (hazard ratio 1.30 [1.02-1.64] for each log10 mcfDNA, p=0.03). Interpretation: High circulating levels of mcfDNA in a substantial proportion of patients with COVID-19 without clinical suspicion for SI suggest that SIs may often remain undiagnosed. McfDNA-Seq, when clinically available, can offer a non-invasive diagnostic tool for pathogen identification, with prognostic value on host inflammatory response and clinical outcomes.