Prediction model based on gut microbiota as a non-invasive tool for gastric cancer diagnosis

Gastric cancer (GC) is a malignant cancer of the digestive tract with high morbidity and mortality. Previous studies have shown that current diagnostic methods largely rely on invasive procedures. Moreover, there are no highly sensitive and accurate biomarkers available for early GC diagnosis. Recent studies using 16S rRNA technology show that gut microbiota can differentiate between diseased and healthy individuals. However, fewer studies emphasize the gut microbiome's value in GC diagnosis. In this study, we collected 455 fecal samples, including 100 from healthy individuals (healthy controls [HCs]), 153 from GC patients, 43 from patients with non-neoplastic diseases of the stomach, and 159 from verification individuals. Our analysis revealed a significantly increased microbial richness in the GC group (Chao1 index, P < 0.05) and distinct compositional differences (principal coordinates analysis). Linear discriminant analysis effect size analysis identified 19 HC-enriched genera (e.g., Bacteroides) and 31 GC-enriched genera (e.g., Streptococcus). The random forest model selected 20 key diagnostic genera, achieving an area under the receiver operating characteristic curve (AUC) of 0.81. By integrating 10 tumor biomarkers, the combined diagnostic model improved the AUC to 0.86 (validation set: 0.84). Tumor biomarker positivity (60.78%) did not directly correlate with microbiota, but the microbiota-biomarker model improved non-invasive diagnostic accuracy, providing a new approach for early GC screening. KEY POINTS: • Changchang Chen and Chen Chen contributed equally to this work • Gut microbiota changes significantly in gastric cancer • Microbiome shows promise as non-invasive diagnostic markers • The combined microbiota-tumor marker model improves diagnosis.