Lung microbiome patterns track with malignancy site and tumor biology
A prospective surgical cohort has added a new layer to the biology of lung cancer: the microbial ecosystem inside resected lung tissue appears to vary with where the malignancy originated and, in some cases, with the tumor’s molecular profile. In 83 patients undergoing lung resection for confirmed malignancy, investigators sampled both bronchial swabs and parenchymal lung biopsies during surgery and sequenced the 16S rRNA gene to profile the local microbiome. The signal was not uniform across every tumor type or every clinical feature, but it was strong enough to suggest that the lung microbiome may be more than passive background noise in thoracic oncology.
The cohort was dominated by primary lung tumors: 75 of 83 patients, or 90.4%, had primary disease, while 8 patients had metastatic lesions. Operations were mostly standard anatomic resections, including lobectomy in 69.9% of cases, wedge resection in 25.3%, and segmentectomy in 4.8%. Those details matter because resected specimens offer direct access to the intrapulmonary environment at the moment of cancer surgery, when tumor tissue, airway secretions, and surrounding parenchyma can be sampled without the contamination risk that can distort sputum-based or noninvasive assays.
Distinct microbiome composition emerged by primary versus metastatic disease
The clearest association involved the broad category of malignancy site. Lung microbiome composition differed between primary lung cancer and metastatic lesions, with a reported p value of 0.014. In clinical terms, that suggests the microbial community in the lung may not simply reflect the presence of cancer in general; it may reflect the biological context of the tumor itself. A metastatic deposit arriving from another organ could interact with the pulmonary microenvironment differently than a tumor that began in the bronchial epithelium or alveolar tissue.
One sentence captures the core signal: the microbial signature was not interchangeable across cancer origins.
That pattern is biologically plausible. Primary lung tumors arise in tissue already exposed to inhaled particles, repeated airway inflammation, local immune surveillance, and the continuous microbial input of the respiratory tract. Metastatic lesions, by contrast, enter an established lung environment from another tissue lineage and may not reshape the local ecosystem in the same way. Even without proving causation, the data point toward a bidirectional relationship in which tumor origin and microenvironment help define one another.
TP53 and KRAS associations suggest a genomic-microbial interface
Tumor mutation status also appeared to align with the lung microbiome. Composition differed by TP53 mutation status, with a p value of 0.040, and showed a near-significant association with KRAS mutations, with a p value of 0.054. These are not trivial genes in lung oncology. TP53 disruption and KRAS activation both mark biologically aggressive, genomically distinct subsets of disease, and both can alter inflammatory signaling, epithelial integrity, and the local immune milieu.
The finding raises a practical question for future work: are the microbial shifts a consequence of tumor-driven inflammation, or could certain microbial communities favor an environment in which specific genomic alterations are selected or sustained? The preprint cannot answer that, but it does show that tumor genomics and the lung microbiome are not isolated compartments. They seem to occupy the same biologic conversation.
No difference in microbiome composition emerged by histology among patients with primary lung cancer. That is an important boundary condition. It suggests that the signal may be more tightly linked to molecular and anatomic context than to histologic subtype alone. In other words, adenocarcinoma versus other primary histologies did not separate the microbiome the way disease origin and selected mutations did.
Prolonged air leaks pointed toward alpha diversity changes
A smaller but clinically meaningful observation involved postoperative recovery. There was a near-significant association between prolonged air leaks and alpha diversity of the lung microbiome, with a p value of 0.070. Although this missed conventional statistical significance, the direction of the relationship is worth attention because prolonged air leak remains one of the more frustrating complications after lung resection, often prolonging chest tube dependence, hospital stay, and recovery.
A diverse microbial community may reflect a more disrupted airway and parenchymal environment, or it may be a marker of tissue characteristics that predispose to impaired healing. Either way, the signal points toward a potential perioperative role for microbiome profiling if future studies confirm the association. Surgical complications are rarely caused by one factor alone; they often emerge from the intersection of tissue quality, airflow dynamics, healing capacity, and inflammation. The microbiome may be another piece of that architecture.
Why surgical sampling matters more than speculation
The study design strengthens the credibility of the findings. Sampling occurred during lung resection, and both bronchial swabs and parenchymal biopsies were analyzed using 16S rRNA gene amplicon sequencing focused on the V4 region. An established bioinformatics pipeline was used to identify taxa and compare microbiome patterns with clinical and tumor variables. That approach matters because the lung has historically been considered a low-biomass site, where contamination and sampling method can overwhelm the true biological signal. Direct operative sampling offers a cleaner window into the tissue microenvironment than many earlier studies could provide.
The work also fits a broader clinical need. Lung cancer care increasingly depends on integrating pathology, genomics, imaging, and perioperative risk assessment. A reproducible microbiome signature could eventually contribute to that framework in several ways: identifying tumors with distinct biologic behavior, refining prognostic models, or helping explain postoperative healing differences. Those uses remain hypothetical, but the current data justify taking the microbiome seriously as a translational variable rather than a curiosity.
Preprint status keeps the findings provisional
This is still a medRxiv preprint and has not undergone peer review. The cohort is relatively small, especially for the metastatic subgroup, and several of the reported associations are modest in statistical strength. The p value for KRAS hovered just above the conventional threshold, and the air leak signal did not cross it. Those limitations do not erase the findings, but they do narrow how far they can be extended.
Validation in larger surgical cohorts will need to show whether these microbiome signatures reproduce across institutions, sequencing workflows, and patient populations. It will also need to determine whether the observed differences are stable enough to support diagnostic or prognostic use, or whether they simply reflect transient shifts in tissue ecology around resected tumors. Until then, the strongest claim is also the most careful one: lung cancer, tumor genomics, and the lung microbiome appear to be linked in a measurable way, and operative sampling may be the right tool to study that relationship.