The Lung Cancer Microbiota and Adjacent Healthy Lung Tissue Microbiota

Background: The microbiota of lung tumor tissue remains incompletely understood. Inflammation related to lung malignancy may correlate with changes in the lung tissue microbiota. We undertook the present study to describe the lung tumor tissue microbiota in relationship to the adjacent lung microbiota.

Methods: 21 matched pairs of lung tissue samples were obtained per Lung Tissue Research Consortium procedures. Each pair of tissue samples consisted of a “tumor” sample obtained from within the tumor or tumor edge and a “non-tumor” sample obtained from unaffected lung at least 5 cm from the tumor. Samples were flash-frozen in liquid nitrogen and maintained at -80°C until DNA extraction using tissue pulverization and mechanical and chemical bacterial lysis. Negative control samples were included to evaluate for contamination. 16S rRNA qPCR was performed to quantify total 16S abundance. The 16S rRNA V4 region was sequenced on the Illumina MiSeq at the University of Minnesota. Human sequences and contaminating OTUs were removed from the dataset. Sequence filtering, denoising, chimera removal, α and β diversity calculations, principal coordinate analysis (PCoA), and taxonomic identification using the Greengenes database were accomplished with QIIME. Statistical analyses were carried out in R.

Results: PCoA demonstrated separate clustering of negative control and tissue samples. After all negative control OTUs were removed from the tissue dataset, 39 samples remained. Using paired t tests, no difference in 16S qPCR copy numbers or Shannon or Simpson a diversity scores between tissue types were found. The most common taxa found among all tissue samples were Lactobacillus, Enterobacteriaceae (family), and Clostridium. No clustering by tissue type was observed on PCoA. After controlling the false discovery rate at 10%, two taxa (Corynebacterium and Peptoniphilus) were significantly more abundant in “tumor” samples than “non-tumor” samples. The correlation between matched pairs of tissue samples was variable, with a mean correlation of 0.15 and a median correlation of 0.031, indicating a small amount of intra-subject correlation in selected subjects.

Conclusion: In our small data set, we did not detect a characteristic lung tumor microbiota.