Back

Plasma Microbial Cell-Free DNA Metagenomic Sequencing Bridges Gaps in the Diagnosis, Epidemiology and Surveillance of Legionella Infections

Nolte, F. S.; Lindner, M. S.; Venkatasubrahmanyam, S.; Pickens, C.; Lim, L.; Hsu, V. P.; Park, S. Y.; Perkins, B. A.; Wunderink, R. G.

2026-03-26 infectious diseases
10.64898/2026.03.23.26348694 medRxiv
Show abstract

Conventional diagnostic methods (CDM) for Legionella preferentially detect L. pneumophila and frequently fail to identify non-pneumophila species (NPLS), obscuring the full clinical spectrum of infection and limiting surveillance accuracy. We analyzed plasma microbial cell-free DNA (mcfDNA) sequencing detections of Legionella spp. from a large clinical cohort tested between 2018 and 2024 and compared species distributions with culture and PCR confirmed cases reported in the most recent national surveillance datasets (2018-2021). To contextualize the clinical impact, we reviewed published reports in which mcfDNA sequencing was used to diagnose legionellosis (2021-2025) and evaluated real-world performance data from a hospital contributing 8.9% of detections within the cohort (Hospital A). mcfDNA sequencing identified proportionally fewer L. pneumophila, more NPLS, and fewer unresolved species than the CDC reports (all p<0.001). Among 15 publications describing 19 U.S. patients, 74% were immunocompromised and 79% had NPLS infections. Concordance between mcfDNA and CDM occurred in 31.6% of cases. At Hospital A with 36 detections, diagnosis was established by CDM alone in none, by both CDM and mcfDNA in 23.5%, and by mcfDNA alone in 76.5%, yielding an additive diagnostic value of 56.8% These findings suggest that plasma mcfDNA sequencing may improve detection of NPLS particularly in high-risk or diagnostically challenging patients and provide complementary data for both clinical diagnosis and epidemiologic surveillance.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.