A Biology-Driven Framework for Selecting, Engineering, and Validating Physiologically Relevant Multicellular Tumour Models

Physiologically relevant tumour models are increasingly used in cancer research, drug development, and functional precision oncology, yet selection often remains platform-led rather than biology-led. This structured critical narrative review proposes an operational framework linking model selection, engineering, and validation to six interdependent domains: multicellular organization and heterogeneity, extracellular-matrix architecture and mechanics, stromal regulation, vascular and interstitial transport, immune regulation and communication, and metabolic regulation with biochemical gradients. The framework distinguishes mechanism-guided questions, in which a candidate causal mechanism can be prespecified, from discovery-oriented questions, in which the determinant is unknown and candidate functions are tested through nested or factorial configurations before a minimal sufficient explanatory model is defined. Required functions are classified as essential, context-modifying, non-essential, or uncertain, and preservation-oriented, reconstruction-oriented, and hybrid strategies are distinguished from the technological format used to implement them. For pharmaceutical applications, delivered exposure, matrix or device binding, tissue penetration, transport, dosing schedule, and recovery are treated as qualification variables when they can change apparent potency or response. The practical contribution is a traceable biology-to-decision record linking the decision and mechanism status to function selection, model configuration, qualification evidence, complexity-retention rules, and claim boundaries. Model qualification is separated from external association, prospective predictive validity, and clinical utility. Among the representative studies appraised, direct functional and configuration-specific pharmacological evidence was demonstrated more often than interlaboratory transfer, independently validated prospective prediction, or clinical utility. Physiological relevance should therefore be demonstrated through decision-critical function rather than inferred from platform complexity or component count. The proposed decision rules remain unvalidated prospectively and across laboratories.