Diabetes mellitus and cancer are two expanding global health burdens that share upstream determinants, yet diabetes may also contribute to malignancy risk through hyperglycaemia/AGE-related stress, insulin resistance with compensatory hyperinsulinaemia, obesity-related inflammation, and tumour microenvironment modulation. This narrative review synthesizes mechanistic plausibility and critically appraises the highest-level clinical evidence on the oncologic safety signals of glucose-lowering therapies, with a focus on cancer incidence captured within randomized diabetes trials. argeted searches of PubMed/MEDLINE and Embase were complemented by manual reference screening. Mechanistic and translational data were summarized separately from clinical evidence. For each drug class, we prioritized pivotal randomized controlled trials and cardiovascular outcome trials (phase 3–4, large, practice-defining), extracting malignancy reporting collected as adverse events/serious adverse events or events of special interest. When randomized evidence was limited, high-quality meta-analyses and selected real-world studies were used for context, clearly distinguished from prespecified randomized evidence. Metformin shows a predominantly neutral effect on cancer incidence, with possible indirect protective signals but no trials specifically designed to assess chemoprevention. Thiazolidinediones demonstrate an overall neutral malignancy profile, with early concerns regarding bladder cancer not confirmed in adjudicated analyses. GLP-1 receptor agonists and dual GLP-1/GIP agonists consistently show no increase in overall or site-specific cancer risk, with emerging signals of reduced obesity-related malignancies. SGLT2 inhibitors display class-wide oncologic neutrality, with heterogeneous and inconclusive drug-specific patterns in meta-analyses. Randomized evidence confirms that insulin, including long-acting analogues, does not increase cancer risk. Current randomized evidence supports a predominantly neutral effect of antidiabetic therapies on cancer incidence, with some agents showing promising signals of protection. This interpretation is constrained by non-prespecified endpoints, limited oncologic adjudication, short follow-up relative to tumour latency, low site-specific event counts, and potential competing-risk effects. Dedicated long-term trials and registry-linkage strategies with cancer-focused endpoints are needed to clarify whether any therapies confer true protection or warrant targeted monitoring.
Modica et al. (Mon,) studied this question.