Abstract Background Luminal B breast cancer is considered immunologically “cold”, with low pathological complete response (pCR) rates after neoadjuvant chemotherapy (NACT). The phase II Neo-CheckRay trial (NCT03875573) tested immune-priming strategies by combining NACT with stereotactic body radiotherapy (SBRT) and immune checkpoint (ICB)/adenosine blockade. Adding durvalumab (arm 2) ± oleclumab (arm 3) to SBRT+NACT (arm 1) doubled pCR rates (from 16.7% to 33.3% and 31.1%). We performed single-cell RNA-sequencing of longitudinal samples from patients included in the Neo-CheckRay trial to elucidate mechanisms of action and identify immune correlates of pCR. Methods We profiled 78 tumor and axillary lymph node (LN) samples at baseline, week 6 (post-SBRT), and at surgery from 28 patients treated at Institut Jules Bordet, using 10X Genomics 5′ platform. Following sequencing and quality control, cells were batch-corrected and annotated using canonical markers. Wilcoxon tests were used to assess changes in cell fractions and gene expression. Results In ICB-treated arms (2+3), high baseline CXCL13+ myeloid cells (p = 0.034) and IL2RA+ T cells (p = 0.002) predicted pCR. Additionally, Arm 3 showed a unique vascular adenosine priming with high CD39/ADORA2A (p = 0.036) on endothelial cells and low ADORA2B (p = 0.017) on endothelial cells.By week 6, SBRT+NACT acted as an in situ vaccine, promoting antigen presentation and T cell priming. Migratory (mDCs, p = 0.002) and conventional DCs (cDC1s, p = 0.02), proliferative myeloids (p = 0.002), early-activated (p = 0.023) and follicular-helper CD4+ T cells (p = 0.006) increased in lymph nodes, while immunosuppressive lipid-associated macrophages decreased (SPP1+TREM2+ LAM p = 0.005, APOC1-LAM p = 0.04). In tumors, arm 1 showed broad immune contraction, T cell exhaustion, and a drop in the effector memory compartment (p = 0.021), while arm 3 displayed rebalanced immunity, with reduced Tregs (p = 0.034) and an increased cytotoxic-to-Treg ratio. ICB responders maintained CXCL13+ myeloids (p = 0.016) and showed enhanced antigen presentation and T cell recruitment via cGAS-STING and IFN signaling, with elevated CXCL10 expression in myeloids (p = 0.018). In arm 3, reduced T-cell exhaustion scores on pericytes (p = 0.013), lower immunosuppression in T cells (p = 0.038), higher MHC II scores on epithelial cells (p = 0.024), and an M1-like myeloid signature indicated successful immune activation.Most strikingly, we observed divergent adenosine pathway remodeling by response and treatment. Arm 3 responders showed redistribution of CD73/CD39 from immune/tumor cells to pericytes/vasculature (p = 0.036, p = 0.01), coupled with increased ADA/ADK expression on immune cells (p = 0.01), enabling active adenosine degradation. In contrast, non-responders and arm 1 accumulated CD73 on myeloids (p = 0.009), CAFs (p = 0.05), and epithelium (p = 0.024) while lacking degradation enzymes, maintaining suppression. Conclusion In immune-cold luminal B breast cancer, effective therapy likely requires: pre-existing immune organization (CXCL13+ myeloid), anti-tumor priming by SBRT+NACT, sustained T-cell activity via PD-L1 blockade, and relief of adenosine-mediated suppression via CD73 blockade. Notably, our data suggest oleclumab’s efficacy may stem not from mere CD73 inhibition, but from its redistribution to the vascular niche, where it may preserve perfusion and immune accessibility, without triggering the immunosuppressive effects seen with CD73 upregulation on myeloid, CAF, or epithelial cells. Citation Format: M. Carausu, A. De Caluwé, D. Gacquer, D. Venet, S. Majjaj, Z. Denis, X. Wang, F. Libert, A. Lefort, I. Desmoulins, A. Gombos, D. T’Kint De Roodenbeke, P. Aftimos, F. Lebrun, D. De Valeriola, I. Veys, C. Pop, S. Drisis, D. Larsimont, M. Ignatiadis, M. Piccart, F. Rothé, E. Romano, L. Buisseret, C. Sotiriou. Heating up cold tumors: single-cell mapping of immune and adenosine pathway reprogramming in luminal B breast cancer (Neo-CheckRay trial) abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr GS2-04.
Carausu et al. (Tue,) studied this question.