The application of bacterial metagenomic analysis as a biomarker for cancer detection is emerging. Our aim was to discover gut microbiota signatures with potential utility in the diagnosis of colorectal cancer (CRC) and non-small cell lung cancer (NSCLC). A prospective study was performed on a total of 77 fecal samples from CRC and NSCLC patients and controls. DNA from stool was analyzed for bacterial genomic sequencing using the Ion Torrent™ technology. Bioinformatic analysis was performed using the QIIME2 pipeline. We applied logistic regression to adjust for differences attributable to sex, age, and body mass index, and the diagnostic accuracy of our gut signatures was compared with other previously published results. The feces of patients affected by different tumor types, such as CRC and NSCLC, showed a differential intestinal microbiota profile. After adjusting for confounders, Parvimonas (OR = 53.3), Gemella (OR = 6.01), Eisenbergiella (OR = 5.35), Peptostreptococcus (OR = 9.42), Lactobacillus (OR = 6.72), Salmonella (OR = 5.44), and Fusobacterium (OR = 78.9) remained significantly associated with the risk of CRC. Two genera from the Ruminococcaceae family, DTU089 (OR = 20.1) and an uncharacterized genus (OR = 160.1), were associated with the risk of NSCLC. Our two panels had better diagnostic capacity for CRC (AUC = 0.840) and NSLC (AUC = 0.747) compared to the application of two other published panels to our population. Thus, we propose a gut bacteria panel for each cancer type and show its potential application in cancer diagnosis.
Keywords: biomarker; colorectal cancer; microbiota; non-small cell lung cancer.