Macrophage Determines Immnophenotype and Predicts Anti-Pd-l1 Response of Urothelial Cancer: Results from Phase II Clinical Trial.

e15093 Background: Immune-checkpoint blockades (ICBs) are revolutionizing treatment of advanced or metastatic urothelial cancer (mUC) whereas the promising response merely occurred in a subset of patients. To dig into underlying determinants, we investigated the comprehensive tumor microenvironment (TME) of 348 tumor samples derived from a large phase II trial (IMvigor210: NCT#02108652). Methods: Based on the 348 pre-treatment samples, the LASSO-Bootstraping algorithm were applied to figure out the optimal predictive biomarker. Assessment of immune-cell proportions, pathway activity, and tumor purity used CIBERSORT algorithm, ssGSEA methodology, and ESTIMATE algorithm, respectively. Unsupervised clustering methods were applied to identify TME infiltrating patterns. Moreover, the predictive value of M1 infiltration in mUC were systematically explored, as well as its correlation to integrative molecular subtypes, tumor intrinsic mutation and metabolic pathways. An additional TCGA dataset was applied to validate the results. Results: Bootstrapping and ROC curve analysis indicated that M1 was non-inferior to TMB in predicting ICBs response (M1: AUC = 0.706; TMB: AUC = 0.728; Delong test: P = 0.333), but surpassed more-than-10,000 counterparts covering TME-infiltrating cells, metabolic pathways, tumor intrinsic pathways and hallmarks of cancer. Kaplan-Meier survival analysis (M1: P < 0.0001, Hazard Ratio = 0.25, 95% CI: 0.14 − 0.44) supported its promising prognostic value. Moreover, study into TME patterns demonstrated its role in determining immunophenotypes (AUC = 0.785). M1 frequency is positively correlated with CD8+ T cells fraction (P = 7e-04), whereas the Treg cell, neutrophil and monocyte infiltration which refer to suppressive immune microenvironment were negatively correlated (P = 0.0066, P = 7e-04, P = 2e-04, respectively). Furthermore, significant correlation was observed between M1 and tumor intrinsic characteristics, including TMB, neoantigen and PD-L1 level of tumor (P = 3.83e-5, P = 1.69e-7, and P = 0.0018, respectively). The TCGA data reproductively supported the majority of aforementioned results. Conclusions: Our study lend statistical power to the notion that Macrophage independently contribute to anti-tumor immune function in the context of ICB therapy. Nonetheless, these findings provide important insight into the role of M1-Macrophage, macrophage polarization and innate immune machinery in therapeutic responses to ICB. Clinical trial information: NCT02951767 .