Tumor immune microenvironment dynamics and outcomes of prognosis in non-muscle-invasive bladder cancer

Abstract

Agents that target PD-1 and PD-L1 have been developed in the treatment of bladder cancer (BC). However, the diversity of immune cell infiltration in non-muscle-invasive BC (NMIBC) and the dynamics of the microenvironment as it progresses to muscle-invasive/metastatic disease remains unknown. To assess tumor immune activity, hierarchical clustering was applied to 159 BC samples based on cellular positivity for the defined immune cellular markers (CD3/CD4/CD8/FOXP3/CD20/PD-1/PD-L1/LAG3/TIGIT), divided into two clusters. There was a "hot cluster" (25%) consisting of patients with a high expression of these markers and a "cold cluster" (75%) comprising those without. The expression of CD39, CD44, CD68, CD163, IDO1, and Ki67 was significantly higher in tumors in the hot cluster. Immunologically, high-grade T1 tumors were significantly hotter, whereas tumors that had progressed to muscle invasion turned cold. However, a certain number of high-grade NMIBC patients were in the cold cluster, and these patients had a significantly higher risk of disease progression. Using an externally available TCGA dataset, RB1 and TP53 alterations were more frequently observed in TCGA hot cluster; rather FGFR3, KDM6A, and KMT2A alterations were common in TCGA cold/intermediate cluster. Analyses of recurrent tumors after BCG therapy revealed that tumor immune activity was widely maintained before and after treatment, and high FGFR3 expression was detected after recurrence in tumors initially classified into the cold cluster. Collectively, we revealed the dynamics of the tumor microenvironment in BC as a whole and identified candidate molecules as therapeutic targets for recurrent NMIBC, e.g., after BCG therapy.

Keywords: FGFR3; PD‐1; PD‐L1; bladder cancer; microenvironment.

FIGURE 1

Representative images of components of tumor immune activity. (A–I) Human bladder cancer tumors immunolabeled for CD3 (A), CD4 (B), CD8 (C), CD20 (D), FOXP3 (E), PD‐1 (F), PD‐L1 (G), LAG3 (H), and TIGIT (I).

FIGURE 2

Overall tumor immune activity in bladder cancer. A hierarchical clustering heatmap (low, blue; high, red) based on the expression levels of the indicated immune cell markers in tumors divided into two clusters, a “hot cluster” and “cold cluster.” The lower panel shows comparisons of other immune‐related molecules between the two clusters.

FIGURE 3

Tumor immune microenvironment dynamics and outcomes in bladder cancer. (A) Percent distribution of the inferred cluster types within each grade and stage of bladder cancer. (B, C) Kaplan–Meier survival curves for progression‐free survival (B) and cancer‐specific survival (C) following total cystectomy in patients with MIBC based on the inferred cluster types. (D–G) Kaplan–Meier survival curves for bladder recurrence‐free survival (D) and progression‐free survival (E) following TURBT in patients with NMIBC. Kaplan–Meier survival curves for bladder recurrence‐free survival (F) and progression‐free survival (G) following TURBT in cold cluster patients with NMIBC divided into two groups; low‐grade tumors in the cold cluster (LG cold) and high‐grade tumors in the cold cluster (HG cold).

FIGURE 4

Genetic alterations in TCGA dataset analysis underlying tumor immune activity. (A) A hierarchical clustering heatmap (low, blue; high, red) based on mRNA expression levels of the indicated immune cell markers in TCGA dataset divided into three clusters: (A) “hot cluster,” “cold cluster,” and “intermediate cluster.” The lower panel shows comparisons of the mRNA expression levels of other immune‐related molecules between the three clusters. (B) Percentage of samples with genomic alterations related to the development of bladder cancer in three clusters.

FIGURE 5

Tumor immune microenvironment in tumors relapsing after BCG therapy. (A) Flow chart of analyses of recurrent tumors after BCG therapy. (B–D) Expression levels for CD3 (B), CD8 (C), and FOXP3 (D) in the two initially classified clusters. (E) Correlation coefficients for PD‐L1 expression levels in samples before and after BCG therapy. (F) Recurrent bladder cancer tumors immunolabeled for FGFR3. (G) Percent distribution of FGFR3 expression (high or low) in recurrent tumors based on initially classified clusters.