Prostate cancer does not advance as a single, uniform disease. Within the same tumor, some glands may retain relatively organized architecture while neighboring regions acquire the molecular features of aggressive malignancy. A new study published in Genes & Diseases uses spatial transcriptomics to trace these changes across prostate cancer tissues, linking the molecular state of individual glandular epithelial regions to Gleason score, pathological stage, and tumor progression. The work identifies SLC4A4 and H2AFJ as particularly promising markers of increasingly aggressive disease.
The study addresses a longstanding limitation in prostate cancer research. The Gleason grading system remains the central method for estimating tumor aggressiveness, but it is based primarily on tissue architecture viewed under a microscope. Although highly valuable clinically, grading does not fully explain how localized, low-grade lesions evolve into advanced carcinomas or why different areas of one tumor can behave so differently. The researchers therefore sought to map gene activity directly within its original tissue context, preserving the relationship between malignant cells, benign glands, and surrounding microenvironments.
To accomplish this, the team analyzed cryosectioned prostate cancer specimens mounted on Superfrost slides containing a spatial microarray of 4,992 barcoded spots. Each spot measured 55 micrometers in diameter and was spaced 100 micrometers from its neighbors, allowing gene expression to be measured across defined microscopic regions. The Visium spatial transcriptomics workflow captured messenger RNA from these locations, generating read-count matrices that recorded which genes were active and where they were expressed. Unlike conventional bulk sequencing, which averages signals across an entire sample, spatial transcriptomics retains a map of molecular activity across the tissue.
The researchers then applied a series of computational analyses to organize the spatial data. Principal component analysis reduced the complexity of the transcriptomic measurements, while Uniform Manifold Approximation and Projection helped visualize relationships among spatial spots. Louvain clustering grouped spots with similar expression profiles into molecularly distinct regions. These clusters were subsequently compared with pathological assessments, enabling the investigators to associate gene-expression states with specific histological structures, including glandular epithelial regions and other components of the tumor microenvironment.
A key component of the analysis was inferCNV, a computational method that estimates large-scale copy-number alterations from gene-expression data. Cancer cells frequently carry gains and losses of chromosomal material, and these changes can provide evidence of genomic malignancy. In the study, inferCNV helped distinguish molecularly abnormal glandular epithelial regions from less malignant or nonmalignant tissue. The investigators found that the inferred genetic malignancy of particular glandular epithelial clusters closely tracked rising Gleason scores, suggesting that spatially localized molecular abnormalities reflect clinically recognized tumor aggressiveness.
The team also reconstructed possible developmental trajectories within the tumor. Diffusion pseudotime, or DPT, was used to arrange cellular or spatial states along a computational progression from less advanced to more advanced conditions. Partition-based graph abstraction, known as PAGA, provided a complementary view of the relationships among these states and helped identify potential transitions between clusters. Together, the analyses offered a model of how glandular epithelial cells may shift from relatively localized or lower-grade states toward increasingly malignant phenotypes. These trajectories do not represent a direct time-lapse of tumor evolution, but they provide a statistical framework for inferring progression from molecular similarities and differences.
To identify genes associated with this progression, the researchers compared gene-expression clusters that aligned with the inferred Gleason-related developmental patterns. They focused on differentially expressed genes, or DEGs, whose activity changed between stages and that satisfied predefined criteria across the analysis. The investigators then looked for genes consistently dysregulated across 12 prostate cancer samples. This cross-sample comparison was important because tumors vary substantially between patients, and genes detected in only one specimen may reflect individual biology rather than a broadly reproducible progression program.
The resulting gene set was enriched for pathways involved in biogenic amine metabolism, broader amine metabolic processes, and arginine and proline metabolism. These pathways may help cancer cells meet the biosynthetic and energetic demands of rapid growth, although the spatial transcriptomic study primarily identifies associations rather than proving that each pathway directly drives progression. Several recognized prostate cancer markers, including FOLH1, AMACR, and KLK3, appeared among the progression-associated genes, providing an internal validation of the approach. The analysis also highlighted SLC4A4, which encodes a bicarbonate transporter, and H2AFJ, a histone H2A variant, as less established but potentially important indicators of aggressive disease.
The investigators tested these candidates at the protein level using immunohistochemistry on prostate tissues representing a range of Gleason scores and pathological T stages, as well as prostatic intraepithelial neoplasia. The cellular staining indices for SLC4A4 and H2AFJ increased significantly with both higher Gleason grades and more advanced pT stages. H2AFJ was particularly enriched in luminal epithelial gland cells, placing its increased expression in the cellular compartment most directly involved in glandular tumor transformation. TFF3 was also included among the candidate proteins evaluated, providing additional validation of the spatially identified molecular changes.
The findings position spatial transcriptomics as a powerful bridge between pathology and molecular oncology. By showing where progression-associated genes are expressed, the approach may help explain why different regions within a prostate tumor carry different levels of risk and could eventually improve tissue sampling, prognostic assessment, and treatment selection. SLC4A4 and H2AFJ are not yet clinical tests, and their biological roles require further investigation through functional experiments and larger patient cohorts. Future studies will need to determine whether these genes actively promote tumor progression, merely reflect aggressive cellular states, or do both. Nevertheless, the work offers a detailed molecular map of prostate cancer development and suggests that metabolic remodeling and epigenetic reprogramming may be central features of the transition toward advanced disease.
Subject of Research: Spatial transcriptomic analysis of prostate cancer progression, with a focus on glandular epithelial cell states, Gleason score, tumor malignancy, and progression-associated biomarkers.
Article Title: “Uncovering genes driving developmental stage progression in prostate cancer through spatial transcriptomics”
Web References: https://doi.org/10.1016/j.gendis.2025.101983; https://www.sciencedirect.com/journal/genes-and-diseases
References: Quan Y, Wang M, Zou F, Zhang H, Zhang Y, Jin Y, Ping H. “Uncovering genes driving developmental stage progression in prostate cancer through spatial transcriptomics.” Genes & Diseases. DOI: 10.1016/j.gendis.2025.101983.
Image Credits: Yongjun Quan, Mingdong Wang, Fan Zou, Hong Zhang, Yishan Zhang, Yongchen Jin, and Hao Ping.
Keywords: Prostate cancer; spatial transcriptomics; Gleason score; glandular epithelial cells; SLC4A4; H2AFJ; inferCNV; tumor progression; cancer metabolism; precision oncology.
