Vascularized tumor organoids enable immunotherapy testing in tumor-remodeled stroma

Accurately predicting whether a new molecule will be safe and effective in humans remains a fundamental challenge in biomedical research, requiring models that can reliably inform clinical outcomes. In oncology, drug attrition rates remain disproportionately high compared to other therapeutic areas, largely due to the poor translatability of conventional preclinical models that fail to capture the complexity of human immune-tumor-stroma interactions (1, 2). Here, we describe human vascularized organoids (VascO), formed through self-assembly of an immune cell perfused capillary network and stromal compartment, interfacing with patient-derived colorectal cancer organoids that enable the combined safety and efficacy assessment of tumor-targeted immunotherapies. VascO generates functional and integrated vascular networks within tumor organoids that sustain organoid growth and recapitulate the morphology and functionality of tumor vasculature. Single-cell transcriptomics reveals that the tumor organoids-educated VascO niche reprograms otherwise healthy fibroblasts into canonical cancer-associated fibroblast states, adopting my-ofibroblastic or inflammatory signatures, and faithfully mirroring donor-specific tumor cues. In parallel, VascO tumor vessels display an expanded tip-cell program, aberrant morphology, and VEGF/ANG2-driven permeability, recapitulating defining hallmarks of tumor vasculature. Immune perfusion with a tumor-targeted T cell bispecific antibody induces dynamic immune trafficking, infiltration and engagement of effector cells with the tumor tissue, leading to its effective killing. A protease-activated variant restricts T cell activation to the tumor microenvironment, retaining potent tumoricidal activity while sparing donor-matched healthy colon organoids, and thereby improving the therapeutic index. VascO thus models a patient-specific, tumor-educated microenvironment that is amenable to assessing the combined safety and efficacy of new drug candidates and provides a scalable platform to improve the preclinical assessment of innovative cancer immunotherapies.