Heterogeneous associations of antidiabetic medications with cancer prognosis: Evidence from 61 studies with over 1.1...

This meta-analysis aimed to comprehensively evaluate the prognostic impacts of seven classes of antidiabetic medications in patients with cancer, predominantly in the setting of concomitant type 2 diabetes mellitus (T2DM), and to elucidate their potential differential effects. A systematic search of PubMed, Embase, Cochrane Library, and the Web of Science was conducted from inception to May 2025, utilizing keywords including "antidiabetic drugs," "cancer," and "prognosis." Based on predefined criteria, eligible English-language randomized controlled trials and cohort studies were included if they compared the prognostic impacts of metformin, insulin, sulfonylureas, dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose cotransporter 2 (SGLT2) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and thiazolidinediones in patients with cancer. The primary outcome domain was all-cause survival, represented by all-cause mortality (ACM) or overall survival (OS) according to the terminology of the original studies. Secondary cancer-related outcomes included cancer-specific mortality (CSM), disease-free survival (DFS), progression-free survival (PFS), and recurrence-free survival (RFS). Multiple independent investigators performed data extraction and quality assessment. Statistical analyses were performed using OnlineMeta V1.1. DerSimonian-Laird random-effects models were used for all primary meta-analyses to calculate hazard ratios (HRs) and 95% confidence intervals (CIs), while fixed-effects estimates were examined as complementary sensitivity analyses. Study quality was assessed using the Newcastle-Ottawa Scale and the Cochrane risk-of-bias tool for randomized trials. A total of 61 studies comprising 1,106,966 patients with cancer were included. Metformin use was associated with lower ACM (HR=0.82, 95% CI: 0.74-0.91, P=0.0001) and CSM (HR=0.77, 95% CI: 0.69-0.87, P<0.0001), whereas insulin use was associated with higher ACM (HR=2.03, 95% CI: 1.63-2.51, P<0.0001). Among the drug classes analyzed, SGLT2 inhibitor use showed a nonsignificant inverse trend with ACM (HR=0.44, 95% CI: 0.11-1.71, P=0.24) and was associated with lower CSM (HR=0.21, 95% CI: 0.20-0.22, P<0.0001), although these estimates were based on few observational studies with study-specific comparator groups and do not establish superiority over other drug classes. In adjusted-HR-only sensitivity analyses, the direction of association was generally consistent for the major metformin outcomes and insulin-related ACM, whereas several other comparisons were attenuated or could not be pooled because too few studies reported adjusted estimates. This meta-analysis identified heterogeneous associations between antidiabetic drug use and cancer-related survival outcomes, with variation across drug classes, cancer types, and study-level mean age groups. Because the evidence was derived predominantly from observational studies with study-specific comparator groups, substantial between-study heterogeneity, and limited data for newer agents, the pooled estimates should not be interpreted as evidence of treatment superiority or used to rank drug classes. These findings are hypothesis-generating and warrant confirmation in prospective studies using clinically comparable treatment groups, standardized outcome definitions, and rigorous control of confounding.