Tongue coating microbiota of patients with gastric cancer

Gastric cancer (GC) is one of the most common cancers, and the non-invasive diagnostic methods for monitoring GC are still lacking. Growing evidence shows that human microbiota has potential value for identifying digestive diseases. We aimed to seek a potential, non-invasive biomarker for GC by using tongue coating microbiota. The tongue coating microbiota was profiled in 181 GC patients and 112 healthy controls (HCs) using 16S rRNA and 18S rRNA genes sequencing technology. Propensity score matching method was used to eliminate lifestyle factors (such as drinking and smoking). Serum metabolomics profiling was performed by ultra-performance liquid chromatography tandem quadrupole time-of-flight mass spectrometry. Random Forest model was constructed for the diagnosis of GC. Linear discriminant analysis effect size (LEfSe) revealed the distinct alteration of tongue coating microbiota between GC patients and HCs in the original population. However, most of the different bacterial taxa remained and most of the different fungal taxa disappeared in the matching population. Some specific HCs-enriched bacterial genera (Porphyromonas, Parvimonas, etc) represented a negative correlation with serum IL-17A and lysophospholipids metabolites in the matching population. By random forest analysis, the combination of six bacterial genera (Peptostreptococcus, Peptococcus, Porphyromonas, Megamonas, Rothia, and Fusobacterium) was the optimal predictive model to distinguish GC patients from HCs effectively, with an area under the curve (AUC) value of 0.85. The model was verified with a high predictive potential (AUC = 0.76 to 0.96). These observations identified that the alteration of tongue coating microbiota had a possible linkage with the inflammations and metabolome, and indicated that tongue coating bacteria could be a potential non-invasive biomarker for diagnosing GC, which might be independent of lifestyle.