Abstract
BACKGROUND/OBJECTIVE: Gastric cancer (GC) is a major cause of cancer mortality worldwide. We evaluated whether oral microbiota could be sensitive, specific, and non-invasive markers for early GC detection. MATERIALS AND METHODS: Saliva samples were analyzed using 16S rRNA sequencing, and oral microbial markers were validated using an internal validation dataset. Machine learning was used to identify key genera, and functional associations were inferred using Kyoto Encyclopedia of Genes and Genomes pathway and ortholog analyses. Blood samples were also collected, and plasma cytokines were quantified by enzyme-linked immunosorbent assay (ELISA) for pathway-level interpretations. RESULTS: Eight genera-Lautropia, Megasphaera, Ralstonia, Pseudomonas, Peptostreptococcus, Anaerovorax, Fusobacterium, and Neisseria-were validated as diagnostic microbial markers (area under the receiver operating characteristic curve [AUC] = 0.91). Megasphaera and Ralstonia were enriched in GC, whereas Lautropia was depleted and associated with reduced risk. These genera may be functionally linked to pathways involved in GC progression, including NF-κB, IL-6, STAT3, TGF-β1, and Smad2/3. The proposed classification method effectively identified early-stage and tumor-marker-negative GCs, underscoring its clinical translation potential. CONCLUSIONS: Oral microbial markers, including Ralstonia, Megasphaera, and Lautropia, may serve as non-invasive diagnostic markers for GC and may be related to carcinogenic signaling activity.