Effects of sodium–glucose cotransporter 2 inhibitors on mortality and heart failure hospitalization in patients with cancer and diabetes mellitus: An updated systematic review and meta-analysis.

312

Background: Cardiovascular disease and cancer frequently coexist, and anticancer therapies can worsen cardiometabolic risk and precipitate heart failure (HF). SGLT2 inhibitors (SGLT2i) reduce HF hospitalization and mortality in diabetic and HF populations, but their benefit in patients with concurrent cancer and diabetes remains poorly characterized. We updated the evidence to evaluate associations between SGLT2i use and all-cause mortality and HF hospitalization in this cardio-oncology population. Methods: PubMed/MEDLINE, Scopus, Embase, and the Cochrane Library were searched from inception through January 23, 2026 (PRISMA-compliant). Observational studies comparing SGLT2i versus non-SGLT2i control in adults with cancer and diabetes were eligible. Primary outcomes were all-cause mortality and HF hospitalization. Risk ratios (RRs) with 95% CIs were pooled using pre-specified random-effects models; common-effect estimates are reported for comparison. Heterogeneity was assessed by I² and τ²; robustness by leave-one-out sensitivity analyses and GOSH diagnostics; small-study effects by funnel plots. Results: Twelve observational studies (n = 144,398; SGLT2i: 49,615; control: 94,783) reported all-cause mortality. SGLT2i use was associated with lower all-cause mortality (random-effects RR 0.50, 95% CI 0.40–0.62), with substantial heterogeneity (I² = 98.4%; τ² = 0.1278). Leave-one-out analyses showed stable estimates across most iterations (RR range 0.40–0.41), except omission of Huang 2023 (n = 50,133), which shifted the pooled RR to 0.60 (95% CI 0.58–0.63) with I² reducing to 87.9%, identifying it as the most influential study. GOSH diagnostics showed distributed heterogeneity across all 4,095 subsets, with no single study as the primary driver. Funnel plot asymmetry was evident, consistent with small-study effects or publication bias. For HF hospitalization, 9 studies (n = 26,694) showed a directionally consistent but statistically inconclusive random-effects estimate (RR 0.57, 95% CI 0.27–1.22; I² = 86.7%), driven partly by two high-variance outliers (Perelman 2024, RR 3.96; Liu 2025, RR 3.58); the common-effect estimate was significant (RR 0.68, 95% CI 0.62–0.75). Leave-one-out analyses under the common-effect model remained consistently below RR 0.75 across all iterations. Conclusions: Across 12 observational studies, SGLT2i use was associated with lower all-cause mortality in patients with cancer and diabetes, though heterogeneity was substantial, funnel plot asymmetry suggests potential publication bias, and confounding by indication cannot be excluded. HF hospitalization showed a directionally consistent reduction that was statistically inconclusive under random-effects modeling.