患者来源肿瘤球体中肿瘤和免疫对抗PD-1及DGK抑制的差异性应答揭示相加性杀伤效应

PD-1 checkpoint blockade has transformed cancer therapy, yet durable responses remain confined to a subset of patients. Diacylglycerol kinase (DGK) inhibition has been proposed as a complementary strategy to enhance T-cell activation and reprogram the tumor microenvironment. We tested a novel dual DGKA and DGKZ inhibitor using a patient-derived organotypic tumor spheroid (PDOTS) platform that preserves tumor, stromal, and immune architecture, enabling functional assessment of therapeutic response in a physiologic ex vivo context. PDOTS from 22 patient tumors were treated with retifanlimab-dlwr (PD-1 inhibitor), DGK inhibitor (INC-DGKi) at two doses, or the combination. Tumor response was defined as a log₂ fold-change < -0.5 in epithelial tumor load (EpCAM⁺) or a composite epithelial marker set to account for squamous cell carcinomas. Immune activation was profiled by multiplexed gene expression emphasizing CD8⁺ effector activity, interferon-stimulated genes, cytokine induction, and antigen presentation. Baseline PD-L1 and MHC-I were measured by flow cytometry. Responsive (R) and non-responsive (NR) tumors were defined by tumor killing and/or immune activation and compared by differential expression and pathway analyses. Retifanlimab induced tumor regression in 6 of 22 (~27%) tumors and triggered immunologic activity in 10 of 22 (~46%), including increased CD8⁺ and NK activity and interferon gene induction. DGKi elicited tumor regression in 5 of 22 (~23%) and stimulated immune-related transcriptional programs in 11 of 22 (~50%), representing largely distinct responders, with pro-inflammatory and leukocyte recruitment pathways enriched. Combination treatment induced tumor regression in 9 of 22 (~41%) and elicited immune activation in 6 of 22 (~27%), exceeding monotherapy cytotoxicity. Baseline PD-L1 and MHC-I variably correlated with response, indicating contextual, not universal, biomarkers. Retifanlimab and DGKi each elicited antitumor and immune responses in distinct tumor subsets, with comparable single-agent activity. Combination therapy improved tumor killing but not immune engagement, suggesting complementarity of the two agents and merits additional studies to clarify the observed discordance between tumor regression and immune signals in the combination setting. These data support biomarker-guided approaches and highlight PDOTS as a translational platform for defining rational immunotherapy combinations.