SGLT-2 inhibitors and heart failure outcomes in cancer patients and survivors: An updated systematic review and meta-analysis.

311

Background: Sodium–glucose cotransporter-2 inhibitors (SGLT-2i) reduce heart failure risk in the general population, but whether that protection extends to cancer patients and survivors is less established. A prior meta-analysis suggested potential benefit; several additional studies have since been published. We performed an updated systematic review and meta-analysis to reassess the association between SGLT-2i use and incident heart failure (HF), HF hospitalization, and all-cause mortality in this population. Methods: We searched PubMed, Embase, the Cochrane Library, and Scopus from inception through 14 January 2026 for studies enrolling cancer patients or survivors that compared SGLT-2i versus non–SGLT-2i control and reported at least one of the three prespecified outcomes. Effect estimates were pooled as risk ratios (RRs) using random-effects and fixed-effect models. Heterogeneity was assessed with I², τ², and Cochran's Q. Robustness was explored through leave-one-out sensitivity analysis and GOSH plots; funnel plots were used to assess small-study effects. Results: Twelve studies were included across outcomes. Incident HF was reported in 4 studies (N=3,073) and was reduced with SGLT-2i (RR 0.39; 95% CI 0.16–0.96; I²=78%). Leave-one-out analysis identified an influential study: omitting Fath 2023 attenuated the estimate to non-significance (RR 0.66; 95% CI 0.41–1.05). GOSH analysis across 15 models showed two distinct clusters — a low-heterogeneity cluster (I²≈0%) centered around RR 0.6–0.7 and a high-heterogeneity cluster (I²=70–90%) centered around RR 0.25–0.4 — consistent with a single influential-study pattern. HF hospitalization was reported in 9 studies (N=9,252) and was lower with SGLT-2i (RR 0.48; 95% CI 0.36–0.66; I²=18.6%). GOSH analysis across 511 models showed the majority of combinations clustering around a protective effect with low-to-moderate heterogeneity, supporting the stability of this estimate. All-cause mortality was reported in 12 studies (N=104,511) and was lower with SGLT-2i (RR 0.47; 95% CI 0.37–0.60), though heterogeneity was very high (I²=98%). GOSH analysis across 4,095 models showed wide dispersion of pooled effects with predominantly high I², though most models remained on the protective side. Conclusions: SGLT-2i use in cancer patients and survivors was associated with lower incident HF, fewer HF hospitalizations, and lower all-cause mortality. The HF hospitalization signal was the most robust, with low heterogeneity and stable pooled estimates across sensitivity analyses. The incident HF result depended substantially on a single study, and the mortality estimate carries little interpretive weight given I²=98%. Prospective studies with prespecified cardio-oncology subgroups are needed before these associations can inform clinical practice.