Abstract

Rationale: Lung cancer is the leading cause of cancer-related death globally, and rising incidence among traditionally low-risk individuals intensifies the need for improved early-detection methods that the lower-airway microbiome may inform. Objectives: To evaluate how respiratory tract sampling method shapes inferred microbiome structure, and whether bronchial brushing recovers a microbial community ecologically distinct from BAL and oral rinse. Methods: Prospective cohort of 33 participants (8 lung cancer, 25 non-cancer controls) underwent oral rinse, bilateral bronchoalveolar lavage (BAL), and bronchial brushing. Microbiome structure was characterized by small subunit ribosomal rRNA gene amplicon sequence variant (ASV) profiling, with indicator-species analysis and SPIEC-EASI correlation-network mapping used to identify ASVs associated with sample type or cancer status. Measurements and Main Results: Sampling method was the dominant axis of variation. BAL communities closely resembled oral rinse (65 jointly indicative ASVs; none shared with brushing), whereas bronchial brushing yielded a distinct but low-biomass signal. After excluding host-derived sequences, two brush-specific indicator ASVs affiliated with Sphingomonadaceae and an uncultured Steroidobacteraceae were identified that co-localized within a single co-occurrence module. Cancer-status effects were not detectable in this pilot cohort, consistent with limited statistical power. Conclusions: Sampling method is the primary determinant of inferred respiratory microbiome structure. Bronchial brushing recovers a distinct but low-biomass signal that is obscured when BAL is used in isolation. However, overlap between this low-biomass signal with host and contaminant sequences indicates that confidently resolving a discrete lower-airway community will require deeper sequencing and dedicated contamination controls. These methodological findings directly inform the design of future cancer- and disease-association studies.