Dados do Resumo

Título

The lung cancer microbiome: an analysis across different sample types reveals new insights for screening methods

Introdução

Lung cancer (LC) is the most prevalent and lethal malignant disease worldwide, and studies have identified important associations between the lung microbiota and the pathogenesis of LC. However, due to the lack of consensus among analysis methods on the microbiome in the context of LC, this can lead to misinterpretation of data and gaps in knowledge of pulmonary oncology. Therefore, evaluating the microbiome under a unified analysis method may provide good insights for more robust diagnostic screenings and targeted therapies for LC.

Objetivo

To describe the lung microbiota in pooled LC samples and compare them with healthy controls.

Métodos

Positive LC samples (N = 408) and Controls (N = 194), both without antibiotic therapy and bronchiectasis, were grouped into: 1. Sputum; 2. Bronchoalveolar lavage fluid (BALF), and; 3. Solid biopsy. NCBI NGS files from four Bioprojects were used for sputum samples, of which three were LC+ (N = 194) and one healthy control LC- (N = 154), four for BALF (N = 76) and four for solid biopsy (N = 138). Using a Shell pipeline, the files were downloaded using SRA Tools, trimmed in Fastp (Phred = 20) and, finally, the microbiome was predicted using the Kraken2 aligner (bank = PlusPFP-16). The taxonomic data analysis and visualization steps occurred via RStudio, in which read normalization (DESeq2 package), diversity estimates (Vegan) and graph plotting (ggplot2/ggVenn) were performed. The statistical tests used were PERMU-Test, for the analysis of principal coordinates, and Kruskal-Wallis followed by Dunnett's test for the multiple comparison of the LC vs. Control groups, since the data did not present a normal distribution.

Resultados

The PCoA plot revealed that sputum samples (LC+), controls and BALF clustered together, while solid biopsy samples clustered separately, showing that the mouth-bronchus intersection is similar in terms of microbial composition in LC, and that there is a significant difference between the variances of the groups (PERMU-Test; p-value < 0.001). The Venn plot showed that there are 52 microbial taxa shared between all groups and the diversity estimate confirmed that the LC+ BALF and solid biopsy groups are statistically different from the LC- Controls (Kruskal-Wallis; p-value = 4.656e-6), while there was no difference between cancer and LC- sputum. The LC+ sputum group showed a greater increase in abundance, in relation to the LC- Controls, of the most predominant bacterial genera, namely: Prevotella, Neisseria, Streptococcus, Veillonella, Haemophilus and Fusobacterium. The score ranked 15 most prevalent microbes, in which Streptococcus took first place in BALF, Sputum and Controls, together with Staphylococcus, in Solid Biopsy. The second position was occupied by Prevotella and Bacillus, while the third, by Veillonella, Pseudomonas and Neisseria.

Conclusões

The study showed that a group of bacterial pathogens that make up the microbiome of sputum samples can serve as a biomarker for lung cancer screening, since they were considerably abundant in relation to the healthy control condition. Furthermore, it was found that most of the bacterial genera mentioned above were present in the ranking of the most abundant taxa in each group, reinforcing the hypothesis that they serve as parameters in clinical practice.

Financiador do resumo

Centro de Computação de Alto Desempenho da Universidade Federal do Pará – CCAD/UFPA
Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – CAPES
Fundação Amazônia de Amparo a Estudos e Pesquisas – FAPESPA

Palavras Chave

lung cancer; microbiome profiling; metagenomics analysis