Background: Metformin is a first-line therapy for type 2 diabetes mellitus and a prominent candidate for oncology drug repurposing following observational associations with lower cancer incidence and mortality in diabetic cohorts. Objective: To synthesise mechanistic rationale and clinical evidence for metformin in oncology, focusing on breast cancer and non-small cell lung cancer (NSCLC), and to identify major sources of heterogeneity and priorities for future research. Methods: Targeted narrative review of peer-reviewed preclinical, translational, and clinical studies, prioritising randomised controlled trials, presurgical biomarker studies, and higher-quality observational analyses with clinically interpretable endpoints. Results: Metformin is proposed to act through host-mediated mechanisms, including improved insulin sensitivity and reduced hyperinsulinaemia and inflammatory signalling, and tumour-cell effects, including mitochondrial complex I inhibition, energetic stress, and AMPK to mTOR modulation. In breast cancer, metformin consistently improves metabolic and inflammatory biomarkers. However, the large MA.32 adjuvant trial in non-diabetic patients showed no overall benefit in invasive disease-free survival or overall survival. Presurgical studies suggest antiproliferative effects such as Ki-67 reduction may be limited to insulin-resistant or otherwise biologically defined subgroups. In NSCLC, evidence is heterogeneous. Combination trials with EGFR tyrosine kinase inhibitors or chemotherapy report mixed efficacy, and gastrointestinal toxicity can limit adherence. Emerging supportive-care studies indicate possible reductions in selected chemotherapy-related toxicities. Conclusions: Current evidence supports context-dependent, rather than universal, oncologic utility of metformin. Future trials should prospectively stratify by metabolic status and tumour biology, optimise dose and timing, and test biomarker-guided combinations and patient-centred supportive endpoints.
Myszewski et al. (Thu,) studied this question.