Association of High Tumor-Stroma Ratio with Prostate Cancer Progression: Insights from Clinical and Genomic Data

Abstract

Background: Tumor stroma ratio (TSR) is a prognostic factor in various cancers, but its role in prostate adenocarcinoma (PRAD) remains unclear. This study investigates TSR's prognostic value in PRAD using clinicopathological data, bulk/single-cell RNA sequencing to explore tumor-stroma interactions and identify therapeutic targets.

Methods: Two PRAD cohorts (The Cancer Genome Atlas cohort, TCGA; Lanzhou University Second Hospital, LUSH) were analyzed for TSR associations with clinicopathological features and biochemical recurrence (BCR). TSR was assessed via digital image analysis and expert pathologist review. Publicly available bulk/single-cell RNA sequencing data were analyzed to identify TSR-associated genes and predict drug targets, pathways, and immunotherapy responses. Quantitative real-time PCR validated mRNA expression. In vitro assays assessed cell proliferation, growth, and migration, while in vivo xenograft assays validated BGN's role in promoting tumorigenesis.

Results: TSR significantly correlated with clinicopathological features (age, Gleason score, stage, seminal vesicle invasion, BCR) in both TCGA (n = 453) and LUSH (n = 320) cohorts. High TSR independently predicted BCR in multivariable Cox regression. High TSR was associated with copy number variations, differentially expressed miRNAs/transcription factors, and metabolic pathways. Predicted anti-cancer drug targets, like Ki8751, showed potential benefit in high-TSR patients. High TSR may correlate with poor immunotherapy response. Notably, downregulation of BGN in cancer-associated fibroblasts (CAFs) significantly suppressed cell proliferation, migration, and invasion in vitro, and in vivo xenograft assays confirmed that BGN downregulation inhibited tumor growth.

Conclusion: This study highlights TSR's prognostic significance in prostate cancer and its association with adverse clinical outcomes and complex tumor-stroma interactions, identifying BGN, a stromal cell-related gene, as a potential therapeutic target for CAFs. However, these findings are limited by the retrospective design, necessitating prospective validation.

Keywords: genomic; prognosis; prostate cancer; single-cell RNA sequencing; tumor-stroma ratio.

Figure 1

The flowchart of this study.

Figure 2

The histological evaluation of tumor stroma ratio (TSR) and immunohistochemistry. (A–D) The results of H&E images were used to divide TCGA and LUSH cohorts into low (A and C) and high (B and D) TSR groups. (E–L) Representative immunohistochemistry images of stromal cells from the LUSH cohort: (1) high expression of androgen receptor (AR, E), Ki67 (F), prostate-specific antigen (PSA, G), and prostate-specific membrane antigen (PSMA, H); (2) low expression of cytokeratin high molecular weight (CK-H, I), p63 (J), P504s (K), synaptophysin (Syn, L). (M) Single cell sequencing data demonstrated the spatial distribution of tumor and stromal cells in 53 prostate cancer patients.

Figure 3

Univariate Cox and multivariate Cox regression analysis of tumor stroma ratio (TSR) groups and clinicopathologic factors in different cohorts. (A and B) Forest plot represents the results of univariate Cox and multivariate Cox regression analysis in The Cancer Genome Atlas (TCGA) and Lanzhou University Second Hospital (LUSH) cohorts. (C and D) Survival curves show biochemical recurrence (BCR) and progression-free interval (PFI) rates of prostate cancer patients with low or high TSR from the TCGA cohort. (E and F) Survival curves show BCR of prostate cancer patients with low or high TSR and Ki67 from the LUSH cohort. (G) The receiver operating characteristic (ROC) curves of risk factors in TCGA cohort. (H) The ROC curves of risk factors in LUSH cohort. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 4

Single-cell landscape of prostate cancer. (A–F) Uniform Manifold Approximation and Projection (UMAP) plots of six datasets (GSE141445, GSE150692, GSE157703, GSE181294, GSE185344, GSE193337) showed single cells colored by the seven cell types (EPCs: epithelial cells; ENCs: endothelial cells; FICs: fibroblast cells; TCs: T cells; BCs: B cells; MACRs: macrophage cells; MACs: mast cells). (G–L) Numbers of overlapping gene pairs in the marker genes of six cell types.

Figure 5

Identification and enrichment analysis of tumor stroma ratio (TSR)-related genes. (A) Spearman correlation analysis was performed to assess the association between cell subpopulation scores or stromal score, immune score, microenvironment scores. (B–E) Levels of stromal score, immune score, microenvironment score, and tumor purity in high and low TSR groups. (F) The Venn diagram shows the hub genes. (G–I) GSVA analysis of GO, KEGG analysis, and hallmark pathway between high and low TSR groups. ***p < 0.001.

Figure 6

Tumor stroma ratio (TSR) groups differ in genomic landscape, drug sensitivity and immunotherapy response. (A) Genome-wide copy number landscape of 460 tumors stratified by TSR groups. Gains (gain + high balanced gain) and losses (loss + high balanced loss) are summarized to the left of the chromosome band panel. (B) A heatmap highlights key miRNA, transcription factor and metabolic pathway differences between high and low TSR groups. (C) Drug response analysis of selected agents in high and low TSR groups based on the Cancer Therapeutics Response Portal (CTRP) and Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) databases. (D) The value of the Tumor Immune Dysfunction and Exclusion (TIDE) from The Cancer Genome Atlas (TCGA) database. (E) Prediction of response to anti-PD1 or anti-CTLA4 immunotherapy by submap in prostate cancer patients between high and low TSR groups. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 7

Prognostic impact of subgroup-associated genes was investigated. (A) Summary of the correlation between expression of top 30 genes with progression-free interval (PFI), biochemical recurrence (BCR), distant metastasis (MET) and overall survival (OS) based on the univariate Cox regression and Kaplan-Meier models. (B–I) Kaplan-Meier survival curves based on the The Cancer Genome Atlas (TCGA) dataset show the association between the expression levels of MYL9, CRYAB, BGN, CRIP2, HOPX, LAPTM5, APOE, and FCER1G genes and biochemical recurrence (BCR) in different subgroups. (J) The expression patterns of these genes were further analyzed in six single-cell datasets (GSE141445, GSE150692, GSE157703, GSE181294, GSE185344, GSE193337), focusing on their expression in corresponding cell subpopulations.

Figure 8

The function of BGN in cancer-associated fibroblasts (CAFs) and the epigenetic regulation of the BGN expression. (A) Validation of differentially expressed mRNAs using real-time quantitative PCR (RT-qPCR). (B) Fluorescence RT-qPCR detection results after siRNA transfection. (C) The Western blotting assay detection results after siRNA transfection. (D–G) Effect of BGN siRNAs on CAFs using CCK-8, colony formation, cell migration, and invasion assays. (H) BGN protein expression in high- and low-TSR prostate cancer tissues by immunohistochemistry. (I) DU145 cell migration in response to conditioned media from BGN siRNAs on CAFs. (J) Photographs of xenograft tumors harvested at the end of the experiment (n = 4), with tumor volumes measured for each group. (K) Comparing promoter methylation levels between cell line based on DepMap dataset. (L) Comparing promoter methylation levels between low or high BGN expression based on TCGA dataset. (M) Correlation between cg04177332 methylation level and mRNA expression level. **p < 0.01, ***p < 0.001.