Metagenomic Next-Generation Sequencing for the Diagnosis of Neonatal Infectious Diseases

Abstract

Infectious diseases pose a fatal risk to neonates. Timely and accurate pathogen detection is crucial for proper clinical diagnosis and therapeutic strategies. Limited sample volumes from neonatal patients seriously hindered the accurate detection of pathogens. Here, we unravel that metagenomic next-generation sequencing (mNGS) of cell-free DNA (cfDNA) and RNA can achieve unbiased detection of trace pathogens from different kinds of body fluid samples and blood samples. We enrolled 168 neonatal patients with suspected infections from whom blood samples (n = 153), cerebrospinal fluid samples (n = 127), and respiratory tract samples (RTSs) (including bronchoalveolar lavage fluids, sputa, and respiratory secretions) (n = 51) were collected and analyzed using mNGS. High rates of positivity (70.2%; 118/168) of mNGS were observed, and the coincidence rate against the final clinical diagnosis in positive mNGS cases reached 68.6% (81/118). The most common causative pathogens were Klebsiella pneumoniae (n = 12), Escherichia coli (n = 12), and Streptococcus pneumoniae (n = 8). mNGS using cfDNA and RNA can identify microbes that cannot be detected by conventional methods in different body fluid and blood samples, and more than 50% of these microbes were identified as causative pathogens. Further local polynomial regression fitting analysis revealed that the best timing for mNGS detection ranged from 1 to 3 days after the start of continuous antimicrobial therapy. Diagnosed and guided by mNGS results, the therapeutic regimens for 86 out of 117 neonatal patients were changed, most of whom (80/86) completely recovered and were discharged, while 44 out of 86 patients completely or partially stopped unnecessary medication. Our findings highlight the importance of mNGS in detecting causative DNA and RNA pathogens in infected neonatal patients. IMPORTANCE To the best of our knowledge, this is the first report on evaluating the performance of mNGS using cfDNA and RNA from body fluid and blood samples for diagnosing neonatal infections. mNGS of RNA and cfDNA can achieve the unbiased detection and identification of trace pathogens from different kinds of neonatal body fluid and blood samples with a high total coincidence rate (226/331; 68.3%) against final clinical diagnoses by sample. The best timing for mNGS detection in neonatal infections ranged from 1 to 3 days, rather than 0 days, after the start of continuous antimicrobial therapy. Our findings highlight the importance of mNGS in detecting causative DNA and RNA pathogens, and the extensive application of mNGS for the diagnosis of neonatal infections can be expected.

Keywords: antimicrobial therapy; infection; mNGS; neonate; the best timing.

FIG 1

Flowchart and pathogen distribution. (a) Participant selection flow diagram. (b) Composition of microbes identified by mNGS. (c) Causative pathogens identified from different types of samples by mNGS. CNS, central nervous system; CSF, cerebrospinal fluid; RTS, respiratory tract sample; n, number of neonates; s, number of samples.

FIG 2

mNGS performance and coincidence rate. (a) Advantages of mNGS over conventional methods. For the samples with positive mNGS (m+) and negative conventional method (C−) results, we further calculated the clinical coincidence rate, and the pie charts show that 50% to 100% of the mNGS results were consistent with the final clinical diagnoses. (b) Agreement of mNGS results with clinical diagnoses. CSF, cerebrospinal fluid; RTS, respiratory tract sample; Cor, coincidence with the final clinical diagnosis; Non-Cor, noncoincidence with the final clinical diagnosis; TCR, total coincidence rate.

FIG 3

mNGS results for cerebrospinal fluid (CSF) and blood samples of neonates with suspected CNS-bloodstream infections. (a) Positivity rates of CSF and blood mNGS tests, TCRs of positive results from both CSF and blood samples, and concordance of pathogens identified in blood and CSF samples. Identical pathogens were identified from the blood and CSF samples of 8 neonates. Different pathogens were identified from the blood and CSF samples of 2 neonates. (b) Pathogen profiles detected in blood and CSF samples.

FIG 4

The positive coincidence rate and the positivity rate of mNGS changed with the duration of empirical therapy before mNGS tests. Green circles represent the positivity rates of mNGS with empirical therapy for 0 days, 0.5 days, 1 day, 2 days, 3 days, 4 to 5 days, 6 to 9 days, and 10 to 15 days. The green curve represents the positivity rates of the mNGS fit using local polynomial regression fitting. The purple circles and curve represent the positivity rates of conventional methods. Bands represent 95% CIs. The blue circles and dashed curve represent the fit positive coincidence rates of mNGS.