PREreview of Implementing rapid pan-microbial metagenomics in paediatric intensive care

Peer Review: Implementing rapid pan-microbial metagenomics in paediatric intensive care

DOI: https://www.medrxiv.org/content/10.1101/2025.10.07.25337257v3 

Background 

This study focuses on evaluating the feasibility, performance, and clinical impact of a rapid respiratory metagenomics service in a pediatric ICU. This study, metagenomics, is a type of genomic sequencing used to analyze respiratory samples in pediatric ICU patients. The researchers adapted a workflow originally developed for adults to detect bacteria, viruses, and fungi in children, enabling rapid identification of pathogens and supporting clinical decisions.

Research Question #1: What is the feasibility, performance, and clinical impact of implementing a previously validated rapid pan-microbial respiratory metagenomics protocol for pathogen detection in pediatric intensive care patients?

• The adapted rapid mNGS protocol showed high specificity and good sensitivity at 16 hours sequencing.

• For bacteria (89%), fungi (100%), and viruses (87%), compared to routine microbiology. mNGS identified 50 additional pathogens in 24% of samples that were missed by routine testing but confirmed by PCR, indicating increased diagnostic yield.

• These were mainly changes in antimicrobial therapy (starting, stopping, or adjusting antibiotics) and some infection control actions.

• Clinical impact showed that 29.3% of samples influenced patient care, with 26.4% leading to antimicrobial changes 

Major Strengths 

•  Novel adaptation and implementation: The study successfully adapted a rapid same-day pan-microbial respiratory metagenomics protocol, previously validated in adults, for use in a pediatric intensive care unit (PICU) setting. 

• This is the first reported clinical service applying rapid metagenomics to pediatric respiratory samples with same-day actionable results, demonstrating feasibility outside the original center and adult population.

• Comprehensive pathogen detection: The protocol detects bacteria, fungi, DNA, and RNA viruses in respiratory samples with good sensitivity and specificity (89% sensitivity for bacteria, 100% for fungi, and 87% for viruses at 16 hours sequencing), expanding diagnostic yield particularly for polymicrobial infections.

• Shows clinical impact and high specificity

• Increased pathogen detection and integration into clinical workflow was shown 

Major Limitations 

• Challenges with antimicrobial resistance detection: The bioinformatics was limited to detecting a restricted set of antimicrobial resistance genes and could not provide comprehensive antimicrobial susceptibility predictions.

• Retrospective assessment without control group: Clinical impact measures were assessed retrospectively without a control, limiting definitive conclusions about patient outcomes or cost-effectiveness.

• Difficulty interpreting polymicrobial samples: High sensitivity detection led to challenges in interpreting low-level presence of common respiratory bacteria in upper respiratory samples

• Sample collection timing constraints: Obtaining samples before 9am for same-day results was challenging, limiting workflow efficiency.

• No discovery of unexpected pathogens: No novel or truly unexpected respiratory pathogens were identified in the cohort

• Limited outcome data: No significant change was observed in overall ICU patient outcomes like length of stay or mortality, though expected for a diagnostic study of this sample size.

Summary

This study represents a significant and successful step in translating rapid metagenomic protocols to pediatric ICU settings with clinically useful and demonstrated impact on antimicrobial management. It provides valuable data on performance metrics and clinical utility, while highlighting challenges related to viral detection sensitivity, antimicrobial resistance prediction, and interpretation of complex respiratory samples. Further studies with cost-effectiveness analyses would better define the clinical benefits and broader applicability of this diagnostic approach.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they used generative AI to come up with new ideas for their review.