Background: The cGAS/STING pathway is an innate immune pathway that promotes cytokine production in response to cytoplasmic DNA, and its activation is important for the induction of anti-tumor immunity. However, predictivity of cGAS/STING tumor expression on the efficacy of PD-1/L1 inhibitors and subsequent immune responses (e.g., changes in serum cytokines) remain to be elucidated. Non-small cell lung cancer with PD-L1 tumor proportion score (TPS) of 50% or higher respond well to immune checkpoint inhibitor (ICI) monotherapy, but there is still a poor response group among them, and the extraction of such patients is an urgent issue. Methods: This is a post hoc analysis of prospective biomarker study, which enrolled 106 patients with advanced non-small cell lung cancer (NSCLC) who were treated with ICI monotherapy between December 2015 and September 2018. We investigated in 68 patients with preserved evaluable tissue samples taken before start of ICI treatment. cGAS, STING, and PD-L1 expression in tumors were stained by immunohistochemistry. cGAS and STING were evaluated by H-score. Using peripheral blood which was collected by the observational study, 41 serum proteins at the time of PD-1/L1 inhibitors initiation and in 4 – 6 weeks later were quantified. Results: The median cGAS and STING H-SCORE were 220 (5 – 300) and 190 (0 – 300), respectively. There were no differences in cGAS or STING H-SCORE between PD-L1 high (TPS ≥50) and low (TPS < 50) groups (p= 0.990 and 0.283). Cases were divided into two groups according to median of the H-SCORE, respectively, and compared. Unexpectedly, cGAS high (H-SCORE ≥220) patients showed significantly shorter progression free survival of ICI when PD-L1 TPS ≥50 (median progression free survival (PFS); 143 days vs. not reached, p = 0.028) and progression free rate at 18 months was 7 and 53%, while no association was observed when PD-L1 TPS < 50 (median PFS; 47 vs. 61 days, p= 0.798). STING tumor expression was not associated with PFS regardless of PD-L1 TPS. In cytokine analysis, cGAS high was associated with significantly higher serum concentrations of TGF-β1 and β2 before ICI initiation (47.5 vs. 22.3hg/l, p= 0.023; 2118 vs. 882rg/ml, p= 0.037), and H-SCORE of cGAS, not STING, were significantly correlated with TGF-β1 and β2 basal levels (r = 0.447, p= 0.009; r = 0.373, p= 0.033). Analysis about fold change from baseline to 4 – 6 weeks later of 41 cytokines revealed that leptin significantly increased in cGAS high tumor while no difference was seen in all of cytokines between STING high and low tumor. Conclusions: Tumor expression (H-Score) of cGAS, not STING, is a candidate for predictor of poor response to ICI monotherapy in NSCLC with PD-L1 high (TPS ≥50). cGAS tumor expression may be associated with TGF-β producing, immune-suppressive tumor microenvironment in NSCLC. Clinical trial information: UMIN000024414.