Background: Immune checkpoint inhibitors (ICBs) for metastatic urothelial carcinoma (mUC) are widely used, but their response rates are inadequate, around 20%. Currently, analyzing samples prior to ICB administration has accumulated so far; thereby predicting response to ICBs by high tumor mutation burden levels and PD-L1 positivity. However, little is known about what kind of tumor microenvironment is formed in recurrent or residual tumors post ICBs, and how it contributes to ICB resistance. Therefore, we applied large datasets of spatial transcriptome analysis to mUC samples post ICBs in order to reveal the diversity of the tumor microenvironment after ICB administration. Methods: The Visium spatial transcriptome analysis was performed on FFPE samples from 12 mUC patients with viable tumor resection after ICB administration. Further, 4 FFPE samples of non-metastatic UC treated-surgically were subjected to spatial transcriptome analysis and were used as ICB-naïve tumor controls. First, the tumor and stromal regions were pathologically annotated. We then identified CD4-positive, CD8A-positive, and FOXP3-positive spots in both tumor and stromal regions, respectively, and analyzed changes in differentially expressed genes (DEGs) and pathways in the tumor space where cancer cells survive after ICB treatment. Results: For DEGs and pathway analysis, 8082 spots in the tumor region and 10964 spots in the stromal region were used. In the tumor and stromal areas, 15.4% and 15.1% were CD4 positive, 6.5% and 6.3% were CD8A positive, and 2.9% and 2.0% were FOXP3 positive, respectively. In tumor regions of ICB-treated specimens, the expression of SULF1, an extracellular matrix-related gene, was increased in all CD4-, CD8A-, and FOXP3-positive spots compared to negative spots. In addition, the expression of complement-related genes such as C1QB, C1QC, and C1R was upregulated in CD8A-positive spots in the tumor regions after ICB. Pathway analysis revealed that oxidative phosphorylation was enriched in CD8A-positive spots in tumor regions prior to ICB, while angiogenesis was enriched in CD8A-positive spots after ICB. Conclusions: Spatial transcriptome analysis of a large dataset of UC samples showed distinct features of immune infiltration involved in therapeutic resistance in the tumor microenvironment under immunotherapy. Notably, T-cells may be exhausted by metabolic changes in CD8A-positive spots in ICB-refractory tumor environment spaces.