Comparative study of plasma microbial cell-free DNA sequencing to culture and polymerase chain reaction in pediatric community-acquired pneumonia with parapneumonic effusion or empyema

TL;DR

Plasma microbial cell-free DNA sequencing shows higher diagnostic accuracy than traditional methods in detecting pathogens in children with complicated pneumonia.

Contribution

The study provides a rigorous real-world evaluation of mcfDNA sequencing against conventional diagnostic methods in pediatric complicated pneumonia.

Findings

mcfDNA sequencing detected pathogens in 86.7% of cases compared to 8.9% for blood culture and 20.0% for pleural fluid culture.

mcfDNA had a positive percent agreement of 91.9% against a composite reference standard of cultures and PCR.

Bacterial cfDNA was detected in 52% of non-bacterial control samples, suggesting potential for false positives.

Abstract

Plasma microbial cell-free DNA (mcfDNA) sequencing is a novel diagnostic tool for pediatric complicated community-acquired pneumonia (cCAP). However, rigorous evaluation of real-world mcfDNA performance is lacking. We compared mcfDNA sequencing to a composite reference standard consisting of blood cultures, pleural fluid (PF) cultures, and targeted PF PCRs in children with cCAP requiring pleural effusion or empyema drainage at Children’s Hospital Colorado from 2022 to 2024. We calculated positive/negative percent agreement (PPA/NPA), Jaccard similarity index, and theoretical time to pathogen detection. We investigated mcfDNA positivity in pediatric controls without bacterial infections. Receiver operating characteristic (ROC) analysis explored potential mcfDNA detection cutoff values to define “true clinical positivity.” Across 45 cCAP cases, mcfDNA sequencing detected a probable pathogen in 86.7% versus 8.9% by blood culture, 20.0% by PF culture, and 71.1% by PF PCRs (P < 0.001). Against the culture+PCR composite reference standard, mcfDNA had a PPA of 91.9% (83.1%–100.0%), NPA of 35.7% (10.6%–60.8%), and a Jaccard index of 0.74. Bacterial cfDNA was detected in 52% of controls. ROC analysis yielded an area under the curve of 0.9, with potential optimal detection cutoffs ranging between 158 and 491 mcfDNA molecules per microliter. Theoretical median time to pathogen detection was 3.0 days with mcfDNA versus 4.5 days for pleural fluid PCRs, driven primarily by PF sampling time. Plasma mcfDNA sequencing had a significantly higher diagnostic yield than cultures and a similar yield to PF PCRs; however, over half of non-bacterial controls had low-level mcfDNA detected, potentially complicating interpretation. mcfDNA detection level cutoffs may help elucidate the clinical significance of detected pathogens.

More sensitive diagnostic tests, particularly non-invasive options, are needed to better identify the causative organism(s) in children with complicated community-acquired pneumonia and help inform pathogen-directed therapy. A novel, potentially powerful diagnostic tool for pneumonia is plasma microbial cell-free DNA sequencing, available commercially as Karius Spectrum. However, unknowns regarding its real-world performance and proper role in clinical practice remain. This study aims to address two ongoing concerns: first, the lack of robust comparisons of microbial cell-free DNA (mcfDNA) sequencing performance against validated conventional and state-of-the-art diagnostic modalities (i.e., pleural fluid testing and blood cultures); and second, the unknown baseline positivity rates of mcfDNA in children without bacterial infections. Results from this study may help inform clinical practice decisions and testing implementation strategies.

Figures