Abstract The long-term benefits of anti-PD-1/PD-L1 are limited in most cancer patients. Resistance and insensitivity are major challenges of this therapeutic class. Feedback regulation of PD-L1 could serve as a mechanism of drug resistant, especially negative feedback loops where inhibition of PD-L1 may lead to up-regulation of its own expression and function. In this study, computational models were established to investigate PD-L1 feedback modulations to facilitate discovery of biomarkers for drug sensitivity and to inform design of drug combination studies. As of October 2018, over 10,000 publications on cancer and immunosuppressive factors (e.g. PD-1, PD-L1, CTLA4, IDO1, TIM-3, LAG3, TIGIT) were found in PubMed. About 2500 abstracts of these articles were downloaded and systematically analyzed through natural language processing and text mining. Information on gene/drug, pathway/disease along with results from cell/animal/clinical studies were extracted, standardized and integrated into a directed graph-based database. Algorithms were designed to construct networks and identify PD-L1 feedback loops (FBLs). Fisher's Exact tests were performed to evaluate the significance of factors involved in (+) and (-)FBLs. The computationally generated PD-L1 feedback network comprises 102 nodes, 216 interactions with 1080 possible FBLs (502 positive vs 578 negative). In particular, up-regulation of IL10, pEGFR, SRC, HIF1A, ISX, NF-kappaB, pSMAD2/SMAD2, IFN-gamma were strongly associated with up-regulation of PD-L1 upon inhibition of PD-1/PD-L1 (p<0.0001, (-)FBLs). Interestingly, TIM-3, a potential IO target and commonly up-regulated by anti-PD-1/PD-L1, is involved in both (+) and (-)FBLs. Inhibition of TIM-3 (e.g. anti-TIM-3) could decrease IL-10 in Tregs and enhance efficacy of anti-PD-1/PD-L1, but meanwhile, anti-TIM-3 could activate NF-kappaB, IRF7 and STAT1, leading to up-regulation of PD-L1 expression and potential resistance to anti-PD-1/PD-L1. In conclusion, the literature-based network analysis proved to be effective to identify potential regulatory pathways driving resistance to immunotherapies, to prioritize the selection of biomarkers, and to assess risk of drug combinations. Particularly, through the analysis of feedback regulations of PD-L1, novel therapeutic approaches to boost efficacy of anti-PD-1/PD-L1 could be postulated. Targeting genes involved in both (+) and (-)FBLs of PD-L1, such as TIM-3, needs to be carefully evaluated. Citation Format: Jian Zhu, Rebecca L. Zhu. Impact of PD-L1 feedback modulation on the sensitivity of anti-PD-1/PD-L1 [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 673.