442 Background: Immune checkpoint inhibitors (ICIs) have revolutionized the therapy pattern of advanced gastric cancer (GC), but the efficacy is still unsatisfactory. The role of metabolism in the immune landscape of tumor microenvironment (TME) has been emphasized for the past few years. As the mediators of cellular interactions in TME, more metabolic biomarkers and their predictive value remain to be revealed. Methods: Metabolite activity scoring gene sets were constructed based on metabolite-protein interaction network. Integrated metabolite profile was collected from metabolic reactions in KEGG database. Transcriptome of ACRG (n=299) and anti-PD-1 cohort (PRJEB25780, n=45) were utilized to generate metabolite activity scores via GSVA algorithm. Cox regression was performed to calculate the hazard ratio (HR) to overall survival of metabolites in each metabolic reaction. Plasma samples (n=55) were obtained to measure nicotinamide metabolites from GC patients in Nanfang Hospital before and after ICIs treatment. Results: 78 metabolic reactions (78/6429) with opposite prognostic value among metabolic substrates and products were identified via the antagonistic metabolism screening workflow. Notably, nicotinamide (NAM)-methylnicotinamide (MNAM) (KEGG REACTION: R01269) showed significant antagonistic relationship (NAM, HR=0.76, 95% CI: 0.64-0.89, P<0.001; MNAM, HR=1.28, 95% CI: 1.09-1.49, P=0.002). NNMT, the rate-limiting enzyme catalyzing the formation of MNAM from NAM, was also correlated with poor prognosis (log-rank P=0.013). In terms of therapeutic prediction, NAM/MNAM ratio in plasma of GC patients was significantly correlated with anti-PD-1 response (PR vs SD/PD, median NAM/MNAM ratio (ng·mL-1/ng·mL-1) = 26.76 vs 10.40, P<0.0001). Compared with baseline, decreased plasma NAM/MNAM ratio was found in all (7/7) PR patients ( P=0.0011), while 73.33% (11/15) of non-responders showed a higher NAM/MNAM ratio after ICIs treatment ( P=0.0283). Correlation analysis of key metabolites with TME cells revealed that MNAM activity score was positively correlated with fibroblasts, MDSCs and M2 macrophages infiltration (Pearson correlation: R=0.60, P<0.001; R=0.42, P<0.001; R=0.34, P<0.001), suggesting that MNAM has a potential impact on immune-suppressive cells recruitment and functions. Conclusions: Nicotinamide metabolism was an indicator of prognosis in GC patients. Circulating NAM/MNAM ratio can be a novel metabolic biomarker in dynamic monitoring for ICIs therapy.