Senescent fibroblasts support prostate cancer treatment resistance, pointing to an AR–p52 target for combination therapy
Background: Treatment resistance also involves the cells surrounding cancer
A study reports that senescence in fibroblasts surrounding prostate tumors contributes to resistance after androgen-deprivation therapy. The researchers identified how NF-κB2/p52 signaling becomes activated in fibroblasts with reduced androgen receptor (AR) levels and reported that inhibiting this pathway could restore sensitivity to antiandrogens in mouse models. The study was published in Theranostics.
Androgen-deprivation therapy (ADT) is a mainstay of prostate cancer treatment, but advanced disease can progress to castration-resistant prostate cancer (CRPC) during treatment. Alongside changes within cancer cells, remodeling of nearby blood vessels, immune cells, and fibroblasts also contributes to resistance. This study examined the microenvironments formed by these cells and sought therapeutic targets in the stroma that supports cancer cell survival.
Key findings: Linking a stroma-centered microenvironment to AR loss
The researchers integrated 10 public datasets containing 133 prostate samples from 103 donors. The samples included healthy and adjacent normal tissue, hormone-sensitive and castration-resistant cancers, and metastatic states of each. After quality control, they analyzed 399,276 cells.
Castration-resistant samples showed “Ecotype 4,” a microenvironment type marked by prominent stromal cells. Spatial analysis and tissue staining found dense clusters of COL4A1-positive capillary endothelial cells and perivascular cells expressing THY1 and POSTN, while CD8-positive T cells appeared to be excluded around these structures. The researchers proposed a model in which TGF-β signaling and collagen accumulation contribute to formation of a vascular–stromal barrier. However, the proposed signaling links between cells include predictions based on transcriptomic data.
An increase in this endothelial cell population was also validated in data from 219 patients with hormone-sensitive prostate cancer and 100 patients with CRPC. In a separate integrated analysis of 1,259 patients, gene-expression signals from this population were associated with early relapse and shorter progression-free survival. These are prognostic associations in patient data, not results from a clinical trial of an inhibitor.
Cancer-associated fibroblasts (CAFs) were central to the mechanistic experiments. Chromatin immunoprecipitation and reporter assays showed that AR binds directly to the promoter of the NFKB2 gene and represses its transcription. Blocking androgen signaling or reducing AR levels led to accumulation of p52, the active form of the NFKB2 gene product. CAFs with increased p52 expression showed higher levels of the senescence marker p16 and senescence-associated secretory factors, including interleukin-6 (IL-6). AR-deficient CAFs also supported the growth of cancer cells and organoids under androgen-depleted conditions.
The researchers targeted this pathway with SN52, a peptide that inhibits p52 nuclear translocation. In organoid co-culture, it weakened the tumor-supporting function of CAFs. In mice, adding SN52 to androgen-blockade conditions that included enzalutamide reduced tumor burden and markers of senescent stroma and IL-6. The main animal experiments testing the combination used six mice per group.
Implications and outlook: A candidate preclinical combination strategy
These findings support a strategy that adds inhibition of stromal p52 signaling to cancer-cell-directed treatment. In particular, the experiments demonstrate a link from AR loss to fibroblast senescence and tumor-supporting activity.
However, efficacy and safety in patients have not been established. It is also unknown whether SN52 can dismantle the vascular barrier or restore immune-cell infiltration in people. The integrated analysis across disease states should not be interpreted as continuous tracking of treatment courses in the same patients. Sample sizes for the animal experiments were not determined by a prospective statistical calculation. Clinical translation will require further validation of the safety of target inhibition, combination efficacy, and the reproducibility of patient-selection markers.
The study design and results are available in the full paper.
Cancer Cell, Published June 2026.
Summary: Bypassing the individual tumor-clone analysis and macroscopic tracking errors that historically compromise conventional therapeutic strategies in metastatic castration-resistant prostate cancer (CRPC), this landmark paper profiles a programmable ecosystem steering infrastructure. Constructing a high-depth single-cell atlas of 399,276 cells from 133 clinical cohorts, the computing platform identifies a convergent therapeutic-refractory macro-state designated Ecotype 4. Mechanistic modeling via CRISPR-Cas9 and ChIP-qPCR established that the therapy-induced loss of the androgen receptor (AR) removes a physiological brake on the non-canonical NF-κB2/p52 signaling pathway, forcing cancer-associated fibroblasts into a highly dense, senescent supportive niche sealed by a rigid, TGF-β-driven vascular-stromal barrier. Pharmacological intervention utilizing the pathway inhibitor SN52 shifted the binding energy matrix, dismantling the immune exclusion boundary to reverse antiandrogen resistance and ensure robust single-cell clinical stratification.
These findings provide a basis for evaluating the state of surrounding fibroblasts alongside cancer-cell responses when studying resistance to antiandrogens. Specifically, researchers could add p52 inhibition in enzalutamide-resistant models and measure tumor growth together with changes in CAF p16 and IL-6 to test how much stromal changes contribute to drug response.
Follow-up studies could also analyze reduced AR levels and senescence markers in patient tissue. It remains unknown whether these markers predict response to combination treatment. There is insufficient evidence to use Ecotype 4 or p52 signaling directly as companion diagnostics, or to use SN52 to treat patients. Separate validation is needed to move from prognostic associations to prediction of treatment response.