According to the fifth edition of the World Health Organization classification, prostate cancer (PCa) encompasses multiple pathological types. However, current single-cell RNA sequencing (scRNA-seq) analyses predominantly focus on adenocarcinomas, with the cellular composition of other rare pathological types remaining largely unexplored. Therefore, our aim is to delineate the single-cell landscape of these rare PCa pathological types, thereby revealing the heterogeneity within tumor microenvironment (TME).

Nine samples from six patients were utilized for scRNA-seq, including prostate adenocarcinoma, mucinous adenocarcinoma of the prostate, prostatic cystadenoma, and treatment-related neuroendocrine prostatic carcinoma. Analyses including InferCNV, Gene Set Variation Analysis, functional enrichment, trajectory, and intercellular interaction assessments were conducted for each identified cell subgroup. Additionally, the correlation and prognostic value of key cell subgroups were calculated using single-sample Gene Set Enrichment Analysis within the The Cancer Genome Atlas (TCGA) database.

We observed heterogeneity both between pathological types and among cell subgroups. We identified a subtype of epithelial cells characterized by epithelial-mesenchymal transition (EMT) features, named LC_KLK4_VIM, which is associated with micrometastases to lymph nodes. Additionally, we revealed that this subgroup interacts with myeloid and T cell subgroups through the MIF-CXCR4 axis, leading to the formation of immunosuppressive TME. Notably, high expression of LC_KLK4_VIM in the TCGA database is correlated with poor prognosis.

Our study delineates the landscape of rare pathological types of PCa, emphasizing the prevalent heterogeneity and revealing that epithelial cells with EMT characteristics are crucial for immunosuppression and disease progression.