Spatially organized immune niches in high-grade primary prostate cancer.

5118 Background: Prostate cancer (PC) has historically been considered immunologically “cold” due to low immune infiltration and a predominantly immunosuppressive tumor microenvironment (TME). However, this understanding is largely based on immune profiling studies of metastatic castration-resistant prostate cancer and may not apply in earlier stages of disease. The composition and organization of the immune TME in primary PC remains poorly defined, especially in relation to Gleason score, the gold standard of pathologic PC grading. We hypothesized that the local PC immune TME is more heterogeneous than previously appreciated, with differences in immune composition and spatial organization between High Gleason (HG) and Low Gleason (LG) PC. Methods: We analyzed 29 radical prostatectomy samples (15 LG vs. 14 HG) using a 40-marker multiplex immunofluorescence panel to profile the TME at single-cell resolution. Immune cell phenotypes were defined with respect to the marker panel. Using R, cell composition was computed within tumor, stroma, and tertiary lymphoid structures (TLS, organized aggregates of immune cells that resemble secondary lymphoid organs) and within defined boundary regions at interfaces between each compartment. Results: We discovered that HG PC exhibited higher total immune infiltration (p < 0.05) and specifically TLS (p < 0.05) relative to LG PC. Furthermore, advancing Gleason grade was associated with increased T cell infiltration (p < 0.05) and M2-like macrophage density (p < 0.05). Strikingly, TLS were enriched in exhausted (PD1+) CD4 and CD8 T cells, especially near Tregs (p < 0.001) and demonstrated a higher M2-like to M1-like macrophage ratio compared to surrounding tumor (p < 0.001) and stroma (p < 0.05). We observed paradoxically increased proportions of CD4PD1 cells (p < 0.05) and Tregs (p < 0.05) in HG vs. LG samples at the tumor-TLS interface. In HG samples, CD8PD1 cells (p < 0.01), CD4PD1 cells (p < 0.05), and Tregs (p < 0.05) were enriched in tumor-TLS interface relative to TLS-stroma interface. HG tumors also demonstrated a higher ratio of M2-like to M1-like macrophages at the tumor-TLS interface compared to intra-tumoral regions (p < 0.05). Conclusions: Our findings challenge the notion that primary PC is uniformly immunologically inert. We show that HG PC has increased levels of immune infiltration. We describe a paradoxical TLS phenotype in HG PC, where TLS harbor high levels of exhausted T cells near Tregs and have an unfavorable macrophage composition. Critically, the tumor-TLS interface emerged as a unique niche of suppression in HG PC, demonstrating multiple features of lymphocytic and myeloid-driven immunosuppression. These immunosuppressive features were not seen in LG PC. The spatial organization of immunosuppressive cells in HG PC provides a rationale for novel immunotherapeutic targeting of the TLS-rich suppressive landscape of HG PC to improve clinical outcomes in the neoadjuvant setting.