Time-series prediction and detection of potential pathogens in bloodstream infection using mcfDNA sequencing

Abstract

Introduction: Next-generation sequencing of microbial cell free DNA (mcfDNA-seq) has emerged as a promising diagnostic method for blood stream infection (BSI) and offers the potential to detect pathogens before blood culture. However, its application is limited by a lack of clinical validation. Methods: We conducted sequential mcfDNA-seq on blood samples from ICU participants at high risk of BSI due to pneumonia, or intravascular catheterization; and explored whether mcfDNA-seq could diagnose and detect pathogens in advance of blood culture positivity. Blood culture results were used as evaluation criteria. Results: A total of 111 blood samples were collected during the seven days preceding and on the day of onset of 16 BSI episodes from 13 participants. The diagnostic and total predictive sensitivity of mcfDNA-seq were 90% and 87.5%, respectively. The proportion of pathogenic bacteria was relatively high in terms of both diagnosis and prediction. The reads per million of etiologic agents trended upwards in the days approaching the onset of BSI. Discussion: Our work found that mcfDNA-seq has high diagnostic sensitivity and could be used to identify pathogens before the onset of BSI, which could help expand the clinical application of mcfDNA-seq.