Abstract
Bladder cancer (BCa) is heterogeneous in the tumour microenvironment (TME). However, the role of the TME in BCa in modulating the response to immunotherapy has not been fully explored. We therefore analysed fractions of immune cells using CIBERSORTx and clustered BCa into subtypes. We also analyzed weighted correlation networks to generate immunotherapy-related hub genes that we used to construct a prediction model using multivariate Cox and LASSO regression analyses. We found that BCa comprised three subtypes (C1‒C3). The prognosis of the patients was the most favourable and the response rate to anti-programmed death ligand 1 (PD-L1) was the highest in C1 among the three subtypes. Immune cells, including CD8(+), CD4(+) memory activated, and follicular helper T cells, activated NK cells, and M1 macrophages infiltrated the C1 subtype. The C2 subtype was enriched in M0 macrophages and activated mast cells, and the C3 subtype was enriched in B and resting immune cells. Mechanistically, the enhanced immunogenicity of subtypes C1 and C2 correlated positively with a higher response rate, whereas the dysregulated ECM-related pathways in the C2 subtype and glycolytic and fatty acid metabolic pathways in the C3 subtype impaired the responses of patients to anti-PD-L1 therapy. We also constructed a TME-related signature based on 18 genes that performed well in terms of overall survival. In conclusion, we determined prognoses and anti-PD-L1 responses by analysing TME heterogeneity in BCa.