Research Bronchoscopy Standards and the Need for Noninvasive Sampling of the Failing Lungs

To the Editor:

We read, with great interest, “Research Bronchoscopies in Critically Ill Research Participants: An Official American Thoracic Society Workshop Report” (1), which outlined research practice standards for bronchoscopy that serve as the reference methodology for sampling of the lower respiratory tract (LRT). We echo the Workshop Report conclusions that research bronchoscopy in the intensive care unit is valuable, safe, and feasible. However, we further highlight the practical challenges and study design considerations associated with bronchoscopy-based research. In response, we propose that ongoing evaluation of the reliability of noninvasive sampling methodologies has the potential to enhance critical care research utilizing LRT specimens. These alternative methods may offer utility beyond resource-limited settings that was pointed out in the Workshop Report.

First, research bronchoscopy may be deferred in the sickest subpopulations with severe hypoxemia or hemodynamic instability, as even the slightest risks of bronchoscopy may be deemed as unacceptable. In contrast, noninvasive samples, such as endotracheal aspirates (ETAs), are part of routine medical care and can be collected through minimal-risk protocols even in the most unstable patients. Noninvasive methods do not require the presence of a study physician, thus allowing for wider reach of enrollment across more study sites. Research bronchoscopy typically requires dedicated informed consent from legally authorized representatives, a process that can pose logistical challenges, hampering early biospecimen collection in the hyperacute phase of critical illness. Therefore, cohorts that include noninvasive LRT sampling may be less prone to selection biases and more representative of the broad patient populations with acute respiratory failure. Furthermore, noninvasive collections can be repeated ad libitum to follow biological trajectories in the LRT. Approaches such as ETAs, exhaled breath condensates, or fluid analysis from routinely discarded heat moisture exchange filters (2, 3) would allow for repeated or even near-continuous assessment of the LRT and may, thus, augment our understanding of the temporal dynamics of LRT biology beyond cross-sectional research bronchoscopy.

Although there are limitations to noninvasive methods, they can capture many features observed in bronchoalveolar lavage (BAL) samples, including cellular heterogeneity, as well as distinct transcriptional (4) and protein signatures (5). For example, ETAs collected without cellular filtration can reveal heterogeneous LRT cellular populations, including neutrophils (6). ETA protein content may reflect important signals from the distal airspaces, as it contains measurable quantities of surfactant protein D (W.B. and G.D.K., unpublished results) and soluble receptor of advanced glycation end products—proteins associated with the distal airspaces (5)—even if their concentrations in ETA may not directly represent BAL levels. The analytical validity of noninvasive biospecimens remains to be consistently demonstrated, yet there is proof-of-concept evidence that rigorous protocols for noninvasive sputum samples can replicate BAL findings in patients with cystic fibrosis (7). Noninvasive sampling is the guideline-recommended method for pneumonia diagnosis in mechanically ventilated patients (8), and microbiome signatures in ETA samples predict patient-centered outcomes (9). Finally, the coronavirus disease (COVID-19) pandemic highlighted the importance of pragmatic LRT sampling across diverse healthcare systems and practices, because bronchoscopies were severely limited by increased clinical demands and heightened infection prevention precautions during the early phases of the pandemic.

Further work is necessary to compare the benefits and limitations of noninvasive methods to the gold standard of bronchoscopy. We recall that, nearly 50 years ago, the diagnostic value of BAL for pneumonia was established through meticulous comparisons against the reference standard of tissue sampling through invasive biopsies or autopsy. We propose that similar meticulous study of noninvasive methods against the current reference standard of bronchoscopy may also allow for development of scalable, reliable, and generalizable approaches for biological study of the LRT and rapid diagnostics during acute respiratory failure.