Pathogen-oriented mNGS is not equivalent to microbiome profiling: interpreting BALF mNGS diversity in ARDS

Abstract

Gao et al. provide valuable data-linking bronchoalveolar lavage fluid metagenomic next-generation sequencing with inflammatory subphenotypes of acute respiratory distress syndrome. We highlight three issues relevant to interpretation: an apparent inconsistency concerning exclusion of samples with no micro-organisms detected, use of a pathogen-oriented workflow for community-level ecological inference, and the distinction between detection yield and diagnostic performance without an independent reference standard. Clarification of sample eligibility, sensitivity analysis including technically valid samples reported as negative by sequencing, and confirmation that sequencing and bioinformatic procedures are validated for quantitative microbiome analysis would strengthen interpretation.

Dear Editor,

We read with great interest the study by Gao et al. examining bronchoalveolar lavage fluid (BALF) metagenomic next-generation sequencing (mNGS) in 159 patients with acute respiratory distress syndrome (ARDS) classified into hypoinflammatory and hyperinflammatory subphenotypes [1]. The finding of lower mNGS-detectable microbial richness in hyperinflammatory ARDS provides an intriguing link between host inflammatory state and pulmonary microbial ecology. Several methodological points, however, may refine interpretation of these results.

First, clarification of the study population appears important. The Methods and enrollment flowchart state that three patients were excluded, because “no microorganisms [were] detected by mNGS” [1]. Yet among the final 159 patients, the original study reports six mNGS-negative cases: four mNGS-negative/CMT-positive and two negative by both mNGS and conventional microbiological testing (CMT) [1]. This represents an apparent inconsistency, because the distinction between the three excluded patients and the six included mNGS-negative patients is unclear. The issue is relevant, because the principal ecological result was reduced mNGS-detectable microbial richness, measured by Chao1 and ACE. Depending on how “mNGS negativity” was operationally defined and applied during cohort assembly, exclusion of samples with no reported micro-organisms could affect representation of the lower tail of mNGS-detectable microbial richness and thereby potentially alter the distribution of richness estimates. We would welcome clarification of the operational definition used for “no microorganisms detected” and, if possible, a sensitivity analysis including all technically valid BALF samples irrespective of mNGS positivity.

Second, the analytical pipeline appears primarily designed for pathogen detection rather than comprehensive microbiome profiling. After host-read removal, sequences were aligned against a database described as covering 26,270 pathogenic micro-organisms [1]. The study nevertheless identified commensal organisms, suggesting that the database is not restricted to classical pathogens; however, it remains unclear how comprehensively non-pathogenic and low-abundance community members are represented. The key issue is, therefore, not whether mNGS can, in principle, be used for microbiome analysis, but whether the sequencing depth, reference database, taxonomic filtering, reporting thresholds, and contamination-control procedures were designed and validated to support quantitative community-level ecological inference rather than pathogen detection alone. A pathogen-oriented clinical mNGS assay may provide taxonomic information, but its readouts should not automatically be interpreted as quantitative representations of the underlying microbial community unless the analytical workflow has been validated for that purpose. This distinction is particularly relevant, because richness estimates are sensitive to the detection of rare taxa. In low-biomass respiratory samples, reagent and environmental contamination can also disproportionately influence apparent community composition [2, 3]. The authors describe negative-control thresholds appropriate for clinical pathogen reporting, but microbiome-oriented statistical contaminant identification, such as prevalence- or concentration-based approaches, may provide additional reassurance [4].

Third, higher microbial detection should be distinguished from diagnostic performance. mNGS was positive in 96.2% of patients compared with 80.5% for CMT, including 29 mNGS-positive/CMT-negative cases [1]. Importantly, the authors themselves acknowledge that no definitive gold standard was available and, therefore, did not use CMT as a reference for calculating diagnostic performance. We agree with this cautious methodological position. Accordingly, descriptions of mNGS as demonstrating “superior diagnostic performance” may be better framed as higher detection yield and broader taxonomic detection, because detection of microbial nucleic acid does not by itself establish that the detected micro-organism is a clinically causal pathogen.

These considerations do not diminish the value of the study. Rather, they suggest that the most defensible interpretation is that hyperinflammatory ARDS was associated with reduced mNGS-detectable microbial richness under the applied sequencing and bioinformatic workflow. Accordingly, the reported Chao1 and ACE indices may be more appropriately interpreted as measures of mNGS-detectable microbial richness under that workflow rather than as direct quantitative representations of the BALF microbial community. Clarifying sample eligibility, validating community-level analytical procedures, incorporating microbiome-specific contamination control, and distinguishing diagnostic attribution from detection yield would further strengthen the promising connection between ARDS inflammatory phenotypes and pulmonary microbial ecology.

Abbreviations

ACE

Abundance-based coverage estimator

ARDS

Acute respiratory distress syndrome

BALF

Bronchoalveolar lavage fluid

CMT

Conventional microbiological testing

mNGS

Metagenomic next-generation sequencing

Author contributions

L.L. and S.L. contributed equally to the conceptualization of the commentary, methodological appraisal, verification and interpretation of the published results and methods, literature review, and critical revision of the manuscript. L.L. additionally contributed to funding acquisition. C.N. contributed to the conceptualization of the commentary, conducted the literature review and critical appraisal, drafted the manuscript, and supervised the overall intellectual content. All authors contributed to the final interpretation, critically reviewed the manuscript, read and approved the final version, and accept accountability for the work.

Funding

This work was supported by the Health Commission of Sichuan Province Medical Science and Technology Program (Grant No. 25LCYJ34).

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.