Single-Cell Map Traces How Prostate Tumors Evolve to Resist Hormone Therapy

Prostate cancer remains one of the most common malignancies in men worldwide, ranking as the second most prevalent cancer and the fifth leading cause of cancer-related death among men globally. For decades, the cornerstone of treatment for advanced disease has been androgen deprivation therapy, or ADT, a strategy that starves tumors of the male hormones they depend on by suppressing androgen receptor signaling. Most patients initially respond: tumors shrink, symptoms ease, and disease is held in check. Yet resistance almost invariably emerges, and the cancer returns in a lethal form known as castration-resistant prostate cancer, which continues to grow even when androgen levels in the blood have been driven down to castrate levels. Understanding exactly how tumor cells transform from hormone-dependent to hormone-independent has been one of the central puzzles in prostate cancer biology, and a new study now offers one of the most detailed single-cell views of that transformation to date.

Researchers at Fudan University Shanghai Cancer Center, writing in the journal Holistic Integrative Oncology, have assembled a comprehensive transcriptional landscape of prostate cancer progression by re-analyzing single-cell RNA sequencing data from 44 patient samples drawn from three prospective clinical trials. Their meta-cohort captured three clinically distinct stages of the disease: 13 samples of localized, treatment-naïve prostate cancer; 25 samples of hormone-sensitive prostate cancer obtained while patients were actively undergoing androgen deprivation therapy; and 6 samples of metastatic castration-resistant prostate cancer, collected from metastatic lesions in the prostate, lymph nodes, bladder, bone, and liver. The hormone-sensitive samples were particularly valuable, having been obtained 13 to 15 weeks after the initiation of a standardized ADT regimen, offering a snapshot of tumor cells in the act of adapting to therapeutic pressure rather than merely before or after it.

The technical challenge at the heart of the study was to reliably distinguish malignant epithelial cells from the benign cells that surround them within each tumor. The team applied a battery of quality-control filters, excluding cells with more than 25 percent mitochondrial gene expression, cells with fewer than 500 detected genes, and potential doublets exhibiting co-expression of incompatible lineage markers. Epithelial cells were identified by high expression of canonical markers such as EPCAM and KRT18, together with the absence of immune or stromal markers. The researchers then turned to inferCNVpy, a computational tool that infers single-cell copy number variation profiles by comparing each epithelial cell against a reference baseline built from 800 immune cells and 800 stromal cells. Clusters showing large-scale copy number aberrations consistent with genomic instability were classified as malignant, while those lacking such patterns were deemed normal. After these stringent steps, the final analytical dataset comprised 35,365 malignant tumor cells from 39 samples: 13 treatment-naïve, 22 hormone-sensitive, and 4 castration-resistant.

The first major finding concerns what happens inside tumor cells during the early months of ADT. Comparing malignant cells from hormone-sensitive samples with those from localized, untreated tumors, the researchers documented extensive transcriptional remodeling. Canonical androgen-responsive genes, including KLK3, the gene encoding the widely used clinical biomarker prostate-specific antigen, were significantly downregulated, confirming that the therapy was doing precisely what it was designed to do: suppressing androgen receptor signaling. Gene set enrichment analysis showed that the hallmark androgen response pathway was markedly suppressed in tumor cells under ADT, consistent with the known pharmacological effects of the treatment.

But the suppression of androgen signaling was only part of the story. In a finding with potentially far-reaching clinical implications, the team observed that ADT induced the activation of immune-associated transcriptional programs within the malignant cells themselves. Hallmark interferon-alpha and interferon-gamma response pathways were significantly enriched, alongside pathways involved in antigen processing and presentation. Over-representation analysis of the most upregulated genes in hormone-sensitive tumor cells further highlighted biological processes related to leukocyte differentiation, humoral immune response, and inflammatory signaling. In parallel, apoptosis-related pathways were significantly enriched, reflecting therapy-induced cellular stress and diminished survival signaling. Crucially, these immune programs were detected within the tumor cells themselves, independent of immune cell infiltration, indicating a direct transcriptional response of malignant cells to androgen deprivation rather than a secondary effect of a changing tumor microenvironment.

This observation aligns with a growing body of evidence suggesting that hormonal therapy can transiently enhance tumor immunogenicity. Prior work has shown that androgen receptor inhibition can increase the expression of major histocompatibility complex class I molecules on prostate cancer cells, potentially making them more visible to the immune system. The new single-cell data provide tumor-cell-intrinsic molecular support for this idea and carry an important practical implication: the timing of combination strategies that pair ADT with immune checkpoint inhibitors may matter enormously. If androgen deprivation opens a transient window of heightened immune visibility in tumor cells, then delivering immunotherapy during that window, rather than after resistance has fully established itself, could maximize therapeutic benefit. Clinical trials are already exploring this logic in metastatic castration-sensitive disease, and the new findings add molecular granularity to the rationale.

The second major finding shifts the focus to the endgame of the disease. When the researchers compared malignant cells from metastatic castration-resistant samples with those from localized, treatment-naïve tumors, they uncovered a strikingly different transcriptional profile, one dominated by genes involved in protein synthesis, mitochondrial function, and oncogenic signaling. Among the most prominently upregulated genes were ribosomal proteins such as RPL41 and RPSA, mitochondrial-encoded genes including MT-ND1 and MT-ND3, and FOLH1, the gene that encodes prostate-specific membrane antigen, or PSMA, a clinically important diagnostic and therapeutic target in advanced prostate cancer. These changes collectively point to an enhanced biosynthetic and metabolic capacity in castration-resistant tumor cells, a cellular state geared toward rapid growth under hostile conditions.

Pathway-level analysis crystallized this picture around a single dominant theme: MYC. Gene set enrichment analysis identified significant enrichment of MYC target gene sets in castration-resistant tumor cells, while androgen response pathways remained significantly downregulated relative to treatment-naïve samples. Over-representation analysis supported the same conclusion through an independent statistical route. To guard against the possibility that the finding was an artifact of their particular cohort, the researchers validated it in an independent dataset, GSE264573, and again found higher hallmark MYC pathway scores in tumor cells from castration-resistant samples compared with castration-sensitive ones. Together, these converging lines of evidence position MYC-driven transcriptional programs as a central feature of tumor cell adaptation in castration-resistant disease.

The emphasis on MYC is biologically coherent with prior mechanistic studies. In localized prostate cancer, MYC has been reported to cooperate with RAS signaling to promote bone metastasis and castration resistance, and a MYC/RAS co-activation signature has demonstrated predictive value across clinical cohorts. In neuroendocrine prostate cancer, an aggressive variant marked by loss of androgen receptor signaling, MYC activation has been associated with neuroendocrine differentiation, and alisertib, an Aurora-A kinase inhibitor that disrupts N-MYC stabilization, has shown clinical activity in a subset of neuroendocrine prostate cancer patients. Still, the authors are careful to note the limits of translation. MYC is not yet an established clinical biomarker in prostate cancer, its activity is not reliably captured by any single genomic alteration, and further studies are required to validate it as a robust prognostic and predictive tool. Moreover, other pathways, including PI3K/AKT signaling, may also contribute to castration resistance even if they were not prominently enriched in this dataset, reflecting the biological heterogeneity of metastatic disease, in which different tumors may rely on different dominant escape routes. The MYC-associated program identified here likely represents one major transcriptional state within a subset of patients rather than a universal mechanism.

The study also carries a molecular rationale for PSMA-targeted medicine. The significantly higher expression of FOLH1 in castration-resistant tumor cells provides single-cell-level support for the clinical efficacy of PSMA-directed therapies, most notably lutetium-177 PSMA-617 radioligand therapy, which has transformed the management of metastatic castration-resistant disease. The authors acknowledge the limitations inherent in their design: the number of castration-resistant samples was small, a constraint that reflects the genuine difficulty of obtaining metastatic specimens suitable for single-cell analysis, and functional validation will be needed to establish causal relationships between the observed transcriptional states and resistance. The analysis also did not systematically link tumor-cell changes to immune infiltration, clonal expansion, or T-cell receptor dynamics, information not available in the underlying dataset. Even so, by defining a transcriptional roadmap that spans from treatment-naïve disease through early hormonal therapy to full castration resistance, the study highlights two actionable features of tumor cell adaptation: a transient immune activation during ADT that could be exploited with timed immunotherapy combinations, and MYC-driven programs in resistant disease that may point toward new biomarker and drug development strategies aimed at delaying or preventing the emergence of lethal, treatment-resistant prostate cancer.

Subject of Research: Single-cell transcriptomic evolution of prostate cancer cells during androgen deprivation therapy and progression to castration resistance

Article Title: Single-cell analysis reveals tumor cell evolution from ADT-naïve to castration-resistant prostate cancer

Article References: Wei, Y., Wu, J., Wang, B., Gu, W., Qin, X., Dai, B., Lin, G., Gan, H., Wan, F., Zhu, Y., & Ye, D. (2026). Single-cell analysis reveals tumor cell evolution from ADT-naïve to castration-resistant prostate cancer. Holistic Integrative Oncology, 5(1), Article 39. https://doi.org/10.1007/s44178-026-00261-0

Image Credits: AI Generated

DOI: 10.1007/s44178-026-00261-0

Keywords: prostate cancer, single-cell RNA sequencing, androgen deprivation therapy, castration resistance, MYC signaling, PSMA, FOLH1, interferon response, tumor evolution, immunotherapy timing, androgen receptor, metastatic cancer