by Christos Evangelou, MSc, PhD – Medical Writer and Editor
Researchers at the University Hospital Heidelberg and the National Center for Tumor Diseases (NCT) in Germany used digital spatial profiling to analyze where proteins are expressed in tumors from patients with prostate cancer. They found that the spatial profiles of proteins in the tumor microenvironment and how their distribution shifts between the tumor’s core and its outer edge may predict prognosis in patients with prostate cancer.
“We could show in this proof-of-concept study that spatially resolved protein expression information has the potential to function as a prognostic biomarker and could therefore be useful for guiding patient management decisions,” said corresponding author PD Dr. Anette Duensing of the University Hospital Heidelberg.
The study was published in Scientific Reports.
Study Rationale
Duensing explained that predicting the clinical trajectory of prostate cancer is challenging and that current prognostic tools, such as stage, grade, serum biomarkers including PSA, and the presence of pathogenic mutations, do not always accurately predict prognosis.
The presence of intratumoral heterogeneity has been recognized in prostate cancer for decades and is even embedded in the Gleason grading system, which accounts for different architectural patterns within the same specimen. What has been less explored is whether the spatial organization of that heterogeneity can provide prognostic information.
“Intratumoral heterogeneity is a common feature of many cancers, including prostate cancer,” Duensing stated in an interview with Pathology News. “The role of spatial niches — defined topological areas populated by tumor cells with certain phenotypical and functional characteristics — is much less understood. Moreover, whether and to what extent spatial niches are clinically relevant is essentially unknown.”
Methodology
The research team used digital spatial profiling (DSP), a platform developed by Nanostring (now Bruker) that measures protein or RNA expression in user-defined regions of interest (ROIs) within formalin-fixed, paraffin-embedded (FFPE) tissue.
The study included 49 patients with high-risk prostate cancer who had undergone radical prostatectomy at the University Hospital Heidelberg. All tissue samples underwent DSP analysis targeting 46 proteins across four panels: an immune cell profiling panel, and assays covering PI3K/AKT signaling, MAPK signaling, and cell death pathways.
Approximately ten ROIs were analyzed per patient, split between the tumor center and the tumor periphery (the zone directly adjacent to non-malignant prostate tissue), yielding a total of 463 ROIs and 21,298 primary data points.
“Tumor center and periphery were chosen because they represent known spatial niches with a very different microenvironment,” Duensing explained. “In each ROI, the expression of 46 proteins was examined simultaneously, focusing on proteins involved in immune signaling, the PI3K/AKT and MAPK pathways as well as apoptosis regulation.”
Neither Niche Predicts Prognosis Alone
A volcano plot comparing protein expression between the two compartments revealed differences in the expression levels of some proteins, confirming that tumor center and periphery are distinct functional niches. Most differentially expressed proteins were upregulated in the periphery. However, the most significantly upregulated protein in the series, the pro-apoptotic BCL2 family member BAD, was overexpressed in the tumor center, confirming earlier findings of the group.
Unsupervised hierarchical clustering of protein expression data from the tumor center alone identified two patient clusters, but neither correlated with progression-free survival (p=0.77). The same was true when hierarchical clustering was performed on protein expression data from the tumor periphery alone (p=0.39).
The prognostic value of protein expression became significant only when the researchers integrated information from both niches. For each of the 46 proteins and each patient, they calculated the log2-transformed relative expression between the tumor periphery and tumor center. Unsupervised clustering of this integrated dataset separated the 49 patients into two groups.
Patients in cluster 2 had a median progression-free survival of 25.7 months following radical prostatectomy. For patients in cluster 1, the median was 8.6 months. The difference in progression-free survival between the two clusters was statistically significant (hazard ratio [HR], 0.43; 95% CI, 0.22–0.86; log-rank p=0.014).
Univariate Cox regression analysis showed that cluster (HR, 2.3; 95% CI, 1.2–4.6; p=0.016), initial PSA value (HR, 2.4; 95% CI, 1.2–4.9; p=0.017) and metastatic stage (HR, 5.0; 95% CI, 2.1–12; p<0.001) were each correlated with the risk of progression. In a multivariate model, cluster association retained a trend for statistical significance, whereas the PSA level was lost, and metastatic stage retained significance.
Duensing emphasized that the two patient clusters did not correlate with standard prognostic variables. Gleason score and BRCA1/2 or TP53 mutation status did not differ significantly between groups. Principal component analysis confirmed the cluster separation was independent of stage, Gleason score, mutation status, and age.
“When combining the protein expression of both spatial niches, unsupervised hierarchical clustering identified two patient clusters that significantly differed with respect to recurrence-free survival,” Duensing said. “Interestingly, these clusters did not correlate with known prognostic parameters, indicating that spatial expression data confers distinct and unique survival information.”
What the Biology Suggests
Cluster 2 (the group with better outcomes) was enriched for patients whose tumors showed higher peripheral expression of immune-related proteins. Among the top ten differentially expressed proteins distinguishing the two clusters were immune-related genes CD3, CD8, CD4, CD45, CD20, PD-1, and CD56, as well as fibronectin, SMA, and BCL6.
One protein group in the clustering analysis was composed entirely of immune regulators, and another was mainly comprised of activated signaling kinases in the PI3K and MAPK pathways. Proteins downregulated in cluster 2 fell into a separate cluster characterized by apoptosis regulators and cell signaling molecules.
According to Duensing, these patterns suggest that in patients with better outcomes, immune cell infiltration at the tumor periphery may be playing a protective role, and that the spatial distribution of oncogenic signaling activity differs from those who progress more quickly.
Clinical Implications and Future Work
The authors acknowledge that this was a proof-of-concept retrospective study with a small cohort from a single center. The use of pre-specified commercial protein panels constrained the scope of what could be measured, and the costs of DSP instrumentation may be prohibitive for many institutions.
“Our results need to be confirmed in a larger patient cohort,” Duensing said. “Another next step will be to validate our findings by other methods with the goal of identifying a smaller, key panel of proteins that is equally capable of conveying prognostic information.”
The study received financial support from the German Federal Ministry for Economic Affairs and Climate Action, the Dietmar Hopp Stiftung, and the Dr. Rolf M. Schwiete Stiftung.
References
- Schneider F, Böning SH, Antunes BC, et al. Prognostic impact of spatial niches in prostate cancer. Sci Rep. 2026;16(1):2598. Published 2026 Jan 17. doi:10.1038/s41598-026-35720-1
