New sequencing tool boosts detection of lung infection germs

Diagnosing lower respiratory infections (LRIs) is often difficult because conventional diagnostic methods (CDMs) can miss suspected pathogens. In work led by Young Jin Kim, researchers set out to compare the newest CDMs with shotgun metagenomic sequencing (SMS) as a way to improve detection. They examined bronchoalveolar lavage (BAL) fluid samples from 16 patients with pneumonia whose samples already tested positive by various CDMs. Those CDMs included bacterial/fungal cultures, real-time PCR for Mycobacterium tuberculosis, testing for cytomegalovirus, and the BioFire® FilmArray Pneumonia Panel. By directly comparing results from these standard clinical tests with the output of SMS, the team wanted to see whether SMS could pick up the same microbes or find additional organisms that the CDMs missed. The study was designed to assess SMS as a supplementary diagnostic strategy rather than a replacement, aiming to measure how often SMS agreed with CDMs and to identify areas where SMS added value in identifying pathogens in the lower respiratory tract.

The researchers ran 10 Gb of shotgun metagenomic sequencing per sample on the NovaSeq 6000 (Illumina) and aligned the sequence reads against the NCBI RefSeq database. For eukaryotic reads they performed an additional matching step using the internal transcribed spacer (ITS) region of fungi. Antibiotic resistance genes (ARGs) were annotated using the Comprehensive Antibiotic Resistance Database model and assessed with the Resistance Gene Identifier. To decide which SMS-detected microbes were significant, thresholds for relative abundance of SMS reads were applied and concordance with CDM-detected microbes was evaluated. The fraction of microbial reads in each sample was very small, ranging from 0.00002% to 0.04971%. SMS detected corresponding bacterial reads ranging from 2 to 23,869 reads and relative abundances between 0.02% and 87.5%. Eukaryotic reads per sample were 0 to 32, and no fungal alignments were found at the genus level, although Candida species were identified in four samples using ITS. No viral reads were detected. Using the predefined thresholds, SMS found pathogens above threshold in 10 of 16 cases (63%); when subdominant taxa were included the figure rose to 11 of 16 cases (69%). ARGs meeting perfect criteria via the Resistance Gene Identifier were observed in two cases.

This comparison represents the first reported head-to-head look at SMS against conventional diagnostics including the BioFire® FilmArray Pneumonia Panel for lower respiratory infections. The results suggest that SMS can often detect the same bacterial pathogens found by standard tests and can identify additional organisms such as Candida species via ITS that were not aligned at genus level in the primary analysis. At the same time, the very low proportion of microbial reads and the absence of detectable viral reads in these samples highlight current sensitivity limitations. The modest rate of concordance—63% by strict thresholds and 69% including subdominant taxa—shows SMS has promise as a supplementary tool but is not yet a complete replacement for CDMs. The detection of antibiotic resistance genes in two cases indicates another potential benefit of SMS, providing genetic insight that could inform treatment. The authors conclude further work is needed to improve sensitivity and make SMS cost-effective before broad clinical adoption.