Beyond Tumor Recapitulation: A Multidimensional Framework for Cancer Organoid Fidelity

Patient-derived cancer organoids (PDOs) are increasingly used to investigate tumor biology, treatment response, and precision oncology, yet their correspondence to the tumors from which they originate is often summarized using global terms such as “recapitulation” or “fidelity.” Such terminology obscures the fact that tumor–organoid correspondence may differ across genomic, clonal, cellular, spatial, metabolic, organelle, temporal, and therapeutic dimensions. Here, cancer organoid fidelity is reframed as a reference-specific and dimension-specific property organized across three analytically distinguishable but biologically interdependent domains: composition, state, and function. We examine how spatial sampling, tissue processing, culture establishment, matrix and medium composition, oxygen and nutrient availability, passage, and multicellular reconstruction can preserve selected tumor features while reshaping others. This distinction is particularly important for metabolism, mitochondrial and lysosomal function, autophagy and mitophagy, ferroptosis, treatment adaptation, and acquired resistance, for which similar phenotypes may arise through different upstream mechanisms. We introduce Culture-Induced State Distortion (CISD) to describe reproducible culture-associated states that differ from the parental tumor reference and propose the Organoid Fidelity Mapping Framework (OFMF) to classify dimension-specific evidence and define the limits of biological inference. Applying the framework to published colorectal cancer and glioblastoma organoid studies illustrates how validated features can coexist with unassessed dimensions. We also consider how reference availability, experimental resources, and clinical feasibility constrain the assessment and interpretation of fidelity. This framework supports question-driven selection of the minimum sufficient model rather than equating increasing experimental complexity with increasing biological fidelity.