Exploding immune cells may explain why prostate drug therapy fails in some men

For millions of aging men, benign prostatic hyperplasia is an unwelcome but predictable companion of growing older. Histologic evidence of the condition appears in roughly half of men by their sixties and in most men thereafter, and the global burden has more than doubled since 1990, reaching an estimated 94 million affected individuals in 2019. The standard first-line drug class, 5-alpha-reductase inhibitors, shrinks the prostate by lowering dihydrotestosterone and demonstrably reduces the risk of acute urinary retention and surgery. Yet a stubborn subset of patients keeps progressing despite treatment, eventually requiring operative intervention, and conventional predictors such as prostate volume, prostate-specific antigen and symptom scores capture only part of that risk. A new study published in Experimental & Molecular Medicine by researchers at Peking University First Hospital and collaborators now offers a striking mechanistic explanation: the drug itself, acting on the prostate’s epithelial cells, may ignite a chain reaction in which immune cells die explosively, neighboring tissue cells are forced into senescence, and the gland’s lining reprograms itself into a drug-tolerant state.

The research team, led by Zheng Li, Zhenan Zhang, Qiancheng Zhao, Zhenpeng Zhu and senior authors Wei Yu and Qian Zhang, combined spatial transcriptomics, single-cell RNA sequencing, immunohistochemistry in fifty patients, functional cell-culture perturbations and mouse genetics to map what they call a macrophage-orchestrated pyroptosis–senescence–club axis in the human prostate. Spatial transcriptomics integrates fourteen tissue sections from public datasets covering normal prostate, untreated benign prostatic hyperplasia and hyperplasia after 5-alpha-reductase inhibitor therapy, retaining 46,176 tissue-covered spots after quality control. By transferring labels from published single-cell references, the team resolved twelve major cell classes and could ask not just which cells were present but where they sat relative to one another, and which ligand–receptor pairs were firing across those microscopic neighborhoods.

The compositional shifts they documented were substantial. In untreated hyperplasia, smooth muscle cells expanded from 13.9 percent of the mapped tissue to 31.6 percent, myofibroblasts surged from 0.47 to 17.3 percent, and luminal epithelium dominated at nearly 40 percent. After 5-alpha-reductase inhibitor exposure, however, the epithelial landscape tilted: the luminal fraction fell to 34.5 percent while club cells, a secretory lineage marked by SCGB1A1 expression, more than doubled from 5.59 to 13.5 percent. Macrophages and B cells also crept upward. Spatial pseudotime reconstruction, rooted in basal-enriched tissue domains, showed that treated samples bifurcated into two transcriptional arms, one resembling a drug-remodeled state enriched for senescence and chemokine programs, and another that persisted near the untreated terminal state, carrying anti-apoptotic and pro-growth signatures the authors describe as tolerance-like.

The pivotal discovery concerns how macrophages are drawn in and then detonated. Under drug pressure, luminal epithelial cells ramped up secretion of macrophage migration inhibitory factor, or MIF, and the chemokine CXCL12. In Transwell chemotaxis assays, conditioned medium from finasteride-treated epithelial cells boosted macrophage migration roughly 3.6-fold using a prostate cancer-derived line and about 4.1-fold using primary luminal epithelial cells isolated fresh from surgical specimens. Blocking MIF or its receptor CD74, or CXCL12 or its receptor CXCR4, each partially curtailed migration, and dual blockade pushed migration back toward baseline. Critically, only the MIF–CD74 axis, not CXCL12–CXCR4, triggered pyroptosis itself, the inflammatory form of cell death in which the NLRP3 inflammasome activates caspase-1 and the pore-forming protein gasdermin D, or GSDMD, ruptures the cell and spills IL-1β, IL-18 and the danger signal HMGB1 into the tissue.

Patient tissue told the same story. Immunohistochemistry across 25 untreated and 25 independently drug-treated patients showed significantly increased CD68 macrophage staining and GSDMD-N, the cleaved active fragment of gasdermin D, in the treated group, with higher macrophage strata aligning with higher pyroptosis grades. Multiplex immunofluorescence and nearest-neighbor distance analysis confirmed that MIF-positive epithelial regions sat measurably closer to CD74- and GSDMD-positive macrophages than background geometry would predict, and that HMGB1-rich pyroptotic macrophages clustered near AGER-positive epithelium, arguing the signaling hotspots were genuine anatomical structures rather than artifacts of mixed measurement spots.

What happens next in the cascade is the senescence step. Conditioned medium from pyroptotic macrophages increased senescence-associated beta-galactosidase positivity in luminal epithelial cells three- to four-fold, drove a pronounced accumulation of cells in the G1 phase of the cell cycle, halved DNA synthesis as measured by EdU incorporation, and elevated the senescence-associated secretory phenotype, with IL-1α, IL-6, IL-8, TNF and MCP-1 rising two- to four-fold. Neutralizing HMGB1 largely reversed all of these effects, whereas blocking IL-1β or IL-18 did little. Co-immunoprecipitation identified AGER, also known as RAGE, as the functional epithelial receptor for HMGB1, and genetic knockdown of AGER rescued the senescence phenotype while receptor overexpression amplified it in an HMGB1-dependent fashion. In patient tissues, both AGER and the senescence marker p16 were significantly elevated after drug treatment.

The final twist is the lineage reprogramming. When the researchers co-cultured recipient luminal cells with senescent primary luminal cells, the senescent cells paracrinely pushed their neighbors toward a club-like identity, increasing SCGB1A1, SCGB3A1 and lactotransferrin while suppressing the androgen receptor, and simultaneously raising the apoptotic threshold through Bcl-2 upregulation and reduced cleaved caspase-3. Knocking down NOTCH1 in the recipient cells partially reversed this conversion, positioning Notch as a druggable effector node downstream of the senescence signal. Patient biopsies corroborated the shift in vivo, showing increased SCGB1A1, reduced androgen receptor and enhanced Notch staining after 5-alpha-reductase inhibitor therapy. The authors interpret the resulting state as a low-proliferation, high-survival adaptation that blunts the gland-shrinking effect the drugs are meant to achieve.

Causal support came from the mouse model. In a castration/testosterone-induced model of prostatic hyperplasia designed to standardize the androgenic background, 5-alpha-reductase inhibitor treatment reduced the prostate index relative to untreated hyperplasia, as expected, but simultaneously expanded the fraction of pyroptotic macrophages from about 21 percent to 37 percent of live tissue macrophages, as quantified by active caspase-1 staining. When the researchers deleted Gsdmd, the pyroptotic fraction collapsed to under 10 percent, senescence and club-like transcriptional programs were attenuated, and prostate enlargement fell further than with the drug alone. Bulk RNA sequencing of the mouse prostates mirrored the human findings, with drug-treated glands enriching pyroptotic inflammatory, luminal-senescence and club-like secretory signatures, all of which were dampened by gasdermin D loss.

The translational payoff is a 13-gene pyroptosis–macrophage signature, dubbed PM-13 and comprising NLRP3, CASP1, CSF1R, CD68, SOD2, IL27, IRF1, CCL2, IL12B, OAS2, CCL7, NLRP2 and TNF, which stratified resistant from responsive patients in an independent longitudinal cohort with area-under-the-curve values of 0.832 at follow-up and 0.858 at baseline. The authors propose a sequential intervention strategy, upstream first and downstream next: de-ignite the cascade by targeting MIF–CD74, dampen the danger signal by blocking HMGB1–AGER or inhibiting GSDMD, and then reshape the epithelium through moderated Notch tuning. Because HMGB1 is detectable in blood and urine, the framework even suggests pharmacodynamic monitoring. The team is candid about limitations, including the cross-sectional nature of the spatial cohort, redundancy among danger signals such as ATP and S100-family proteins, the partial reversal achieved by Notch inhibition alone, and the fact that the mouse model uses castration plus testosterone rather than directly phenocopying clinical monotherapy. Even so, the study reframes drug resistance in benign prostatic hyperplasia not as a failure of hormone blockade but as a programmable immune–epithelial adaptation, and it hands clinicians four concrete molecular switches with which to intervene before a patient’s prostate outlasts the prescription.

Subject of Research: Macrophage pyroptosis-driven epithelial senescence and lineage reprogramming in benign prostatic hyperplasia treated with 5-alpha-reductase inhibitors

Article Title: Macrophage pyroptosis drives luminal senescence and club-cell reprogramming in 5α-reductase inhibitor-treated prostatic hyperplasia

Article References: Li, Z., Zhang, Z., Zhao, Q., Zhu, Z., Yang, K., Meng, Y., Zhang, Q., & Yu, W. (2026). Macrophage pyroptosis drives luminal senescence and club-cell reprogramming in 5α-reductase inhibitor-treated prostatic hyperplasia. Experimental & Molecular Medicine. https://doi.org/10.1038/s12276-026-01864-3

Image Credits: AI Generated

DOI: 10.1038/s12276-026-01864-3

Keywords: benign prostatic hyperplasia, 5-alpha-reductase inhibitors, macrophage pyroptosis, GSDMD, HMGB1–AGER signaling, MIF–CD74 axis, cellular senescence, club-cell reprogramming, spatial transcriptomics, single-cell RNA sequencing, Notch signaling, PM-13 gene signature