Pulmonary microbiota signatures adjacent to adenocarcinoma, squamous cell carcinoma and benign lesion

Abstract

Introduction: The occurrence and progression of lung cancer are influenced by pulmonary microbiota, yet the relationship between changes in the pulmonary microbiota and lung cancer remains unclear.

Methods: To investigate the correlation between pulmonary microbiota and the signature of lung lesions, we analyzed the microbial composition at sites adjacent to the stage 1 adenocarcinoma, squamous carcinoma and benign lesion tissues in 49 patients by using 16S ribosomal RNA gene sequencing. We then conducted Linear discriminant analysis, receiver operating characteristic (ROC) curve analysis and PICRUSt prediction based on 16S sequencing results.

Results: Overall, the microbiota composition at sites close to lung lesions showed significant differences between different lesion types. Based on the results of LEfSe analysis, Ralstonia, Acinetobacter and Microbacterium are the dominant genera of lung adenocarcinoma (LUAD), lung squamous carcinoma (LUSC) and benign lesions (BENL), respectively. Furthermore, we determined the diagnostic value of the abundance ratio of Ralstonia to Acinetobacter in adenocarcinoma patients through ROC curve analysis. The PICRUSt analysis revealed 15 remarkably different metabolic pathways in these lesion types. In LUAD patients, the increase of the pathway associated with xenobiotic biodegradation may be due to the continuous proliferation of microbe with degradation ability of xenobiotics, which implied that LUAD patients are often exposed to harmful environment.

Discussion: The abundance of Ralstonia was related to the development of lung cancer. By measuring the abundance of microbiota in diseased tissues, we can distinguish between different types of lesions. The differences in pulmonary microbiota between lesion types are significant in understanding the occurrence and development of lung lesions.

Keywords: adenocarcinoma; lung cancer; pulmonary microbiota; ralstonia; squamous cell carcinoma.

Figure 1

α-diversity in bacterial communities in three classifications of lung lesions. Shannon index and Simpson index were indexes to evaluate α-diversity, and there were differences in the LUAD group, BENL group, and LUSC group. *P<0.05, **P<0.01.

Figure 2

(A) At the genus level, the LUSC group and the BENL group had higher similarity and a closer cluster analysis distance, while the LUAD group was far apart and had poor similarity compared with the LUSC and BENL groups. (B) PCoA diagram and PERMANOVA analysis showed that there were significant differences in pulmonary microbial structures between LUAD and BENL groups (P = 0.001), LUAD and LUSC groups (P = 0.002), but no significant differences between LUSC and BENL groups (P = 0.16).

Figure 3

Lung microbiota in lung cancer and relevance. (A) The abundance of Ralstonia in the LUAD group was significantly higher than that of BENL and the LUSC group by statistical analysis of the top three genera. (B) The abundance of Microbacterium in the LUSC group was significantly higher than that in the LUAD and BENL groups. The relative abundance of the displayed microbes was presented in a heat map (−log 10). Different colors corresponded to different abundances; smaller numbers corresponded to higher abundances. (C) The abundance of Acinetobacter in the BENL group was significantly higher than that in the LUSC and LUAD groups.

Figure 4

ROC curves of the three groups of highly abundant strains. (A) Ralstonia (AUC = 0.887) had good specificity and sensitivity to distinguish LUAD and LUSC. (B) Acinetobacter (AUC = 0.8871) has good specificity and sensitivity to distinguish between LUAD and BENL. (C) The prediction effects of Ralstonia (AUC = 0.7388) and Acinetobacter (AUC = 0.7517) were good and had certain diagnostic value. (D) The abundance ratio of Ralstonia to Acinetobacter (AUC = 0.8121) can well predict LUAD and unLUAD.

Figure 5

PICRUST-based lung microbiota study in patients with LUSC, LUAD, and BENL. Results are colored by the lesion’s category and sorted in decreasing order of degree of increase within each category. Note that the LDA score (log 10) was >3.0, and the region 3.0–4.0 displayed pathways with more increase.