Beyond the tumor microenvironment: a hierarchical immune-circuit model of immune checkpoint blockade resistance

Immune checkpoint blockade (ICB) produces durable tumor control in a subset of patients, yet resistance is usually interpreted through tumor-intrinsic lesions or suppression within the tumor microenvironment (TME). We propose that ICB resistance can also be organized as failure of a multiscale antitumor immune circuit comprising local immune execution, regional immune education in tumor-draining lymph nodes (TDLNs), and systemic immune calibration by the host. In this model, "hierarchical" denotes nested functional dependence rather than one-way anatomical control: local killing depends probabilistically on antigen visibility, access, and a renewable supply of tumor-reactive cells; nodal priming depends on competent dendritic-cell licensing and host immune fitness; and reciprocal feedback can propagate or repair failure across compartments. We first define the canonical CD8-dominant substrate that ICB can amplify, including cross-presentation, costimulation, progenitor-exhausted T-cell maintenance, trafficking, and target-cell recognition, while retaining alternative CD4, natural killer, intratumoral antigen-presenting-cell, and tertiary lymphoid routes. We then evaluate local, regional, and systemic resistance mechanisms, four directional feedback axes, and context-dependent patterns in pancreatic cancer, melanoma, non-small-cell lung cancer, microsatellite-stable colorectal cancer, hepatocellular carcinoma, and prostate cancer. Evidence from animal perturbation, human spatial and clonal studies, and clinical trials supports individual circuit components but remains heterogeneous. Randomized perioperative regimens demonstrate disease- and setting-specific benefit, whereas negative randomized results and mixed or limiting early-phase signals across innate agonism, stromal or metabolic targeting, radiotherapy combinations, and systemic conditioning constrain therapeutic extrapolation. We therefore present compartment-resolved biomarkers and adaptive trial designs as research hypotheses, not a validated classifier or standard-care algorithm. The framework will be useful only if prespecified multiscale measurements improve prediction beyond tumor-only models and if mechanism-matched interventions produce the expected pharmacodynamic repair before clinical benefit is attributed to the circuit.