Background <p>Additional surgical resection is required to achieve curative treatment in patients with early gastric cancer (EGC) due to the potential risk for lymph node metastasis (LNM) after pathological analysis; however, LNM is estimated to occur in approximately 10% of patients with high-risk EGC. In this study, we investigated a blood-based liquid biopsy assay of exosomal microRNA (miRNA) for the non-invasive detection of LNM in patients with high-risk EGC.</p> Methods <p>Two genome-wide miRNA expression profiling datasets [GSE164174 and The Cancer Genome Atlas (TCGA)] were analyzed to prioritize biomarkers in pretreatment plasma samples from clinical training and validation cohorts of GC patients. An integrated exosomal miRNA panel was developed and a risk stratification model combining the miRNA panel with clinical risk factors was established.</p> Results <p>Using comprehensive expression profiling of public datasets, we identified a transcriptomic panel of four miRNAs (miR-34b, miR-130a, miR-375, and miR-627) that robustly identified patients with LNM [area under the curve (AUC) 0.86, 95% confidence interval (CI) 0.77–0.92]. We assessed panel performance in a training cohort (AUC 0.86, 95% CI 0.67–0.96) and validated it in an independent validation cohort (AUC 0.83, 95% CI 0.68–0.94). Our risk stratification model was more accurate than the panel and was an independent predictor of LNM identification (AUC 0.94).</p> Conclusions <p>A novel, non-invasive, liquid biopsy-based method for patients with EGC may predict those conventionally classified as high-risk patients with LNM who are unlikely to benefit from surgical resection.</p>