Hsin-Yuan Fang

This study was designed to characterize the microbiomes of the lung tissues of lung cancer patients. RNA-sequencing was performed on lung tumor samples from 49 patients with lung cancer. Metatranscriptomics data were analyzed using SAMSA2 and Kraken2 software. 16S rRNA sequencing was also performed. The heterogeneous cellular landscape and immune repertoires of the lung samples were examined using xCell and TRUST4, respectively. We found that nine bacteria were significantly enriched in the lung tissues of cancer patients, and associated with reduced overall survival (OS). We also found that subjects with mutations in the epidermal growth factor receptor gene were less likely to experience the presence of Pseudomonas. aeruginosa. We found that the presence of CD8+ T-cells, CD4+ naive T-cells, dendritic cells, and CD4+ central memory T cells were associated with a good prognosis, while the presence of pro B-cells was associated with a poor prognosis. Furthermore, high clone numbers were associated with a high ImmuneScore for all immune receptor repertoires. Clone numbers and diversity were significantly higher in unpresented subjects compared to presented subjects. Our results provide insight into the microbiota of human lung cancer, and how its composition is linked to the tumor immune microenvironment, immune receptor repertoires, and OS.

Background: Analyzing fusion gene transcripts may yield an effective approach for selecting cancer treatments. However, few comprehensive analyses of fusions in non-small cell lung cancer (NSCLC) patients have been performed. Methods: We enrolled 54 patients with NSCLC, and performed RNA-sequencing (RNA-Seq). STAR (Spliced Transcripts Alignment to a Reference)-Fusion was used to identify fusions. Results: Of the 218 fusions discovered, 24 had been reported and the rest were novel. Three fusions had the highest occurrence rates. After integrating our gene expression and fusion data, we found that samples harboring fusions containing ASXL1, CACNA1A, EEF1A1, and RET also exhibited increased expression of these genes. We then searched for mutations and fusions in cancer driver genes in each sample and found that nine patients carried both mutations and fusions in cancer driver genes. Furthermore, we found a trend for mutual exclusivity between gene fusions and mutations in the same gene, with the exception of DMD, and we found that EGFR mutations are associated with the number of fusion genes. Finally, we identified kinase gene fusions, and potentially druggable fusions, which may play roles in lung cancer therapy. Conclusion: The clinical use of RNA-Seq for detecting driver fusion genes may play an important role in the treatment of lung cancer.

Analysis of The Cancer Genome Atlas data revealed that alternative splicing (AS) events could serve as prognostic biomarkers in various cancer types. This study examined lung adenocarcinoma (LUAD) tissues for AS and assessed AS events as potential indicators of prognosis in our cohort. RNA sequencing and bioinformatics analysis were performed. We used SUPPA2 to analyze the AS profiles. Using univariate Cox regression analysis, overall survival (OS)-related AS events were identified. Genes relating to the OS-related AS events were imported into Cytoscape, and the CytoHubba application was run. OS-related splicing factors (SFs) were explored using the log-rank test. The relationship between the percent spliced-in value of the OS-related AS events and SF expression was identified by Spearman correlation analysis. We found 1957 OS-related AS events in 1151 genes, and most were protective factors. Alternative first exon splicing was the most frequent type of splicing event. The hub genes in the gene network of the OS-related AS events were FBXW11, FBXL5, KCTD7, UBB and CDC27. The area under the curve of the MIX prediction model was 0.847 for 5-year survival based on seven OS-related AS events. Overexpression of SFs CELF2 and SRSF5 was associated with better OS. We constructed a correlation network between SFs and OS-related AS events. In conclusion, we identified prognostic predictors using AS events that stratified LUAD patients into high- and low-risk groups. The discovery of the splicing networks in this study provides an insight into the underlying mechanisms.