Abstract Background: Prostate cancer (PCa) is the second leading cause of cancer-related death among men, primarily driven by hormonal mechanisms. First-line treatments include androgen deprivation therapy and androgen receptor pathway inhibitors (ARPIs) ; however, resistance to these treatments often leads to castration-resistant prostate cancer. Cancer-associated fibroblasts (CAFs) within the tumor microenvironment (TME) significantly influence treatment resistance; however, their interaction with tumor cells remains poorly understood due to a lack of suitable model systems. Methods: We hypothesized that CAFs reduce the cytotoxic effects of docetaxel chemotherapy and the ARPI darolutamide. To investigate this, we established two 3D co-culture systems: STACKs and LumeNEXT allowing us to examine treatment outcomes in the context of the PCa TME. STACKs enables high-throughput assessments, while LumeNEXT recapitulates the TME architecture, including engineered blood vessel mimics to study vascular and epithelial interactions involved in metastasis. We treated androgen-sensitive (LNCaP, LAPC4) and androgen-resistant (22Rv1) tumor spheroids with 10 μM darolutamide or 20 nM docetaxel, in the presence or absence of immortalized CAFs (hPrCSC-44) or normal myofibroblasts (WPMY-1) within the STACKs system. Cytotoxicity was assessed using confocal microscopy. Ongoing work aims to enhance the model by incorporating patient-derived fibroblasts from PCa patients. We characterized primary fibroblasts through bulk RNA sequencing and histological analysis and optimized co-culture conditions before introducing primary CAFs into the LumeNEXT model. Additionally, we included new genetically engineered tumor cell lines (p53/Rb1 double knockout, wild-type, and APIPC) that represent common mutations found in PCa. To measure cell death, we calculated the ratio of Ethidium homodimer+ and Hoechst+ nucleated cells to the total number of Hoechst+ cells in spheroids (≥100 µm). Statistical analyses were conducted using a two-way ANOVA and Tukey's multiple comparisons test. Results: In the STACKs, CAFs significantly reduced docetaxel toxicity in LAPC4, LNCaP, and 22Rv1 compared to conditions without CAFs (LAPC4: p0. 001, LNCaP: p0. 001, 22Rv1: p0. 001). Similarly, CAFs decreased darolutamide toxicity in LAPC4 and LNCaP cells (both p0. 001), while as expected 22Rv1 showed no response to darolutamide. Myofibroblasts had no significant effect on treatment efficacy. In the LumeNEXT system, CAFs again diminished darolutamide and docetaxel-induced cell death in LNCaP cells compared to conditions without CAFs (Docetaxel: p0. 001, Darolutamide: p0. 001). Conclusions: Our findings indicate that CAFs can attenuate the cytotoxic effects of ARPIs and docetaxel in 3D PCa TME models, highlighting their crucial role in treatment resistance. Ongoing research will further investigate the impact of patient-derived fibroblasts and the underlying molecular mechanisms through single-cell RNA sequencing. Citation Format: Nikolett Lupsa, Erika Heninger, Kate Vietor, Shannon R. Reese, Xavier T. Hazelberg, Jacob Popp, Aaron M. LeBeau, David J. Beebe, Sheena C. Kerr, Joshua M. Lang.: Modeling CAF-mediated therapy resistance in 3D prostate cancer systems using STACKs and LumeNEXT abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Prostate Cancer Research and Treatment; 2026 Jan 20-22; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86 (2Suppl): Abstract nr A042.
Lupsa et al. (Tue,) studied this question.