Beyond RECIST: rethinking endpoints and biomarkers for educative cancer therapy

Non-cytotoxic cancer therapies increasingly seek to control malignant systems rather than maximize tumor-cell killing. Educative strategies aim to impose or restore less harmful states such as dormancy, differentiation, tumor microenvironment normalization or reduced invasion. Pressure-dependent evolutionary containment is a related control strategy, not evidence of durable reprogramming by itself. The limitations of size-based assessment are already well established: Choi, PERCIST, mRECIST, Lugano and iRECIST each operationalized a disease-specific alternative. This review's contribution is therefore narrower than "moving beyond RECIST." We propose an Educative Response Framework that defines admission criteria for durable state-reprogramming claims, with mechanism- and pharmacology-informed tests of persistence and prespecified falsification rules. It separates four evidentiary layers: tumor burden, malignant behavior, clonal evolution and patient-level benefit. Biomarker purpose is described using FDA-NIH BEST terminology, while a proposed evidence scale distinguishes analytical validation, prognostic or predictive validity, and formal surrogate validation; the clinical utility of biomarker-guided management is assessed separately. Endpoints are classified as pharmacodynamic, intermediate or candidate, formally validated surrogate, or direct patient benefit, with each assignment justified for its disease and setting and regulatory use recorded separately. The common measurement layers also apply to adaptive containment without implying that it meets the durability criterion. We examine neoantigen vaccines, matrix metalloproteinase inhibitors and ctDNA-guided adjuvant trials to identify gaps between mechanistic activity and clinical benefit, and propose a validation roadmap. The aim is not to replace RECIST, but to make educative claims falsifiable without overstating what a biomarker or endpoint establishes.