Mechanism-guided drug repurposing against therapeutically actionable resistance domains in cancer: from resistance biology to clinical translation

Tumor resistance remains a central obstacle in cancer treatment. It blunts the activity of otherwise effective agents and narrows the therapeutic window, so that patients may accrue toxicity without commensurate benefit, and it drives progression and relapse. Resistance is rarely a single lesion; it is the emergent product of interacting programs that limit intracellular drug accumulation through ATP-binding cassette (ABC) transporter-mediated efflux, provide extrinsic protection within tumor microenvironment (TME) niches, sustain the survival and plasticity of cancer stem cells (CSCs), and maintain quiescent dormant cells that seed late recurrence. Drug repurposing, the identification of new therapeutic uses for approved or clinically characterized drugs, is attractive in this setting because many agents developed outside oncology act on the transport, metabolic, stromal, inflammatory and stress-adaptive pathways on which these programs depend. In principle, existing human pharmacokinetic, formulation and safety data may shorten development timelines and reduce early attrition relative to de novo discovery, although mechanism confirmation and prospective clinical validation remain obligatory. Metabolic inhibitors, anti-inflammatory agents, antiparasitic and antiviral drugs have all been proposed to disrupt survival mechanisms engaged under therapeutic pressure, yet candidate selection and mechanistic attribution in resistant tumors remain difficult. In this review we apply a single mechanism-first filter to representative candidates in each of the four domains above, separating preclinical signals from clinically validated outcomes and stating explicitly where a conclusion rests on one primary study. We then examine how resistance biology can be linked to biomarker-defined patient selection. Finally, we consider the trial architectures, including basket, umbrella and platform designs, through which repurposed agents can be evaluated without diluting a context-restricted mechanistic signal.