Co-evolution of cancer cells and their microenvironment of primary and lymph node metastatic prostate cancer under androgen deprivation therapy

Prostate cancer (PC) is the most commonly diagnosed malignancy and the second leading cause of cancer-related death among men in the United States. The presence of lymph node (LN) metastases is a strong prognostic indicator of progression. Although LNs are the second most frequent metastatic site after bone, they are the first site of detectable metastasis in patients. Currently, the molecular and microenvironment transitions from primary tumors to LN metastases remain poorly understood. Prior bulk sequencing studies have provided important insights into metastatic tumors but failed to resolve the cellular heterogeneity and spatial organization in the tumor microenvironment (TME). We hypothesized that both cancer cell-intrinsic transcriptional programs and their interactions with the TME drive LN metastasis and resistance to androgen deprivation therapy (ADT). To test this, we integrated single-nucleus RNA sequencing with multiple spatial transcriptomic platforms (10x Visium, Vizgen MERSCOPE, and Xenium) to profile primary prostate tumors and matched LN metastases. Spatial analyses revealed site-specific remodeling of the TME between cancer and adjacent normal regions in both the prostate and LN. Despite being distinct organs, the prostate and LN also exhibit overlapping enrichment of specific immune and stromal cell populations within tumor regions, reflecting conserved functional roles across these environments. Non-negative matrix factorization identified recurrent cancer cell metaprograms (MPs), including an androgen receptor (AR) associated program enriched in LN metastases. By integrating MP activity with TME composition through ecotype analysis, we defined distinct ecotypes representing combinations of cancer states and their microenvironment cell types. One ecotype characterized by AR signaling and protumorigenic TME components was preferentially enriched in LN metastases. Ecotype-based stratification of primary tumors identified a ‘pre-metastatic' subgroup, which showed ecotype contributions similar to those of LN metastases, and a ‘non-metastatic' subgroup, which displayed different ecotype profiles. Primary tumors characterized by pre-metastatic ecotypes were linked to significantly poorer clinical outcomes. Collectively, these findings delineate the co-evolution of prostate cancer cells and their microenvironment during LN metastasis. Our study establishes ecotypes as a framework for linking cancer-intrinsic programs with TME remodeling, providing new biomarkers of progression and resistance and offering opportunities for precision therapeutic intervention.