e15516 Background: Peritoneal metastasis (PM) is the most frequent mode of distant metastasis in advanced gastric cancer (GC). Most PM could cause peritoneal effusion. However, the cytological examination of cancer cells in peritoneal effusion is less sensitive to the identification of PM early; the specificity of serum biomarkers in peritoneal effusion is poor. It is necessary to assess the feasibility of peritoneal effusion circulating tumor DNA (ctDNA) for the PM diagnosis. Methods: 22 GCs with clinically confirmed PM as positive subjects and 11 non-cancers without PM as negative control were recruited. GC tissues were collected from 16 positive subjects. Peritoneal effusion supernatant (AS), peritoneal effusion cells (AC), and plasma were collected from each subject. Next generation sequencing (NGS) was performed using a 1021-gene panel. Sensitivity and specificity for the diagnosis of GC with PM were assessed. Results: Median age of 22 GCs with PM was 57 yrs. 12 patients were male. Diffused and mixed-type accounted for 73%. For 45% of patients, primary tumor was found in gastric body. In tissue samples, recurrent genes were TP53 (63%), CDH1 (50%), RHOA (19%), TGFBR1 (19%), KRAS (13%), and PIK3CA (13%). Private and tumor-derived mutations were found in the AS, AC, and plasma. 95% of (21/22), 82% (18/22), and 77% (17/22) of AS, AC, and plasma samples carried mutations, respectively. AS had more mutations than AC and plasma (Wilcoxon’s matched-pairs signed-ranked test, AS vs AC, p= 0.006; AS vs plasma, p = 0.013). 98% of the AC and 80% plasma mutations were contained in AS, Variant allele frequency (VAF) in AS was higher than others (Mann Whitney test, AS vs AC, p= 0.038; AS vs plasma, p< 0.001). In 22 GCs with PM, 3 were identified with no cancer cells in AS using cytological examination. Of these, 3 were detected with mutations in AS, one with mutations in AC. Mutations were detected in only one negative control AS and AC. Receiver operating characteristic (ROC) curve showed the number of mutations was related to PM (AS: AUC 0.93, 95% CI 0.82-1.00, p< 0.001; AC: AUC 0.85, 95% CI 0.70-1.00, p= 0.001). When the threshold was determined with one mutation, the sensitivity and specificity were 95% and 91% in AS, 82% and 91% in AC, respectively. Conclusions: Peritoneal effusion ctDNA contained tumor-derived and specific mutations. Peritoneal effusion ctDNA can be used for the auxiliary diagnosis of PM.