Summary: In this post-hoc analysis of the EXSCEL randomised trial (4749 participants with type 2 diabetes and a baseline ejection fraction), left ventricular ejection fraction (LVEF) modified the effect of once-weekly exenatide on hospitalisation for heart failure (p-interaction = 0.012). The hazard ratio was 1.52 (95% CI 0.95-2.43) in those with LVEF below 40% and 0.74 (95% CI 0.55-1.01) in those with LVEF of 40% or higher. Neither subgroup estimate was individually statistically significant, so this is a hypothesis-generating signal rather than proven benefit or proven harm.
PICO Summary
| Element | Detail |
|---|---|
| Population | Adults with type 2 diabetes from the EXSCEL trial; n = 4749 of 14,752 had a baseline LVEF available (455 [10%] with LVEF <40%, 4294 [90%] with LVEF ≥40%). Post-hoc analysis of an international, double-blind, placebo-controlled randomised trial. |
| Intervention | Once-weekly subcutaneous exenatide 2 mg (EQW), randomised arm. |
| Comparison | Matching placebo, randomised arm. |
| Outcome | Hospitalisation for heart failure (main outcome). LVEF modified the exenatide effect: HR 1.52 (95% CI 0.95-2.43) for LVEF <40% and HR 0.74 (95% CI 0.55-1.01) for LVEF ≥40%; p-interaction = 0.012. Neither subgroup HR reached significance (both CIs cross 1.0). The effect was also modified by NYHA class (HR 0.91, 95% CI 0.65-1.27 for class I/II; HR 1.84, 95% CI 0.95-3.59 for class III/IV; p-interaction = 0.062, not significant), driven mainly by the LVEF <40% subgroup. No significant treatment-by-LVEF interaction was seen for MACE, cardiovascular death, or all-cause mortality (all p-interaction >0.10). Obesity did not modify the effect. Absolute risk reduction and NNT were not reported for these subgroup analyses. |
Exenatide & heart failure by ejection fraction
RCT post-hoc · type 2 diabetes · EXSCEL
Ejection fraction modified the exenatide effect on heart failure hospitalisation (p-interaction 0.012), but neither subgroup hazard ratio was individually significant. A hypothesis-generating signal, not proof of benefit or harm.
Expert Commentary
This is a secondary, post-hoc analysis of the EXSCEL cardiovascular outcomes trial, and it should be read as exploratory rather than confirmatory. The headline is an interaction, not an effect: ejection fraction appeared to modify how once-weekly exenatide related to heart failure hospitalisation, with a directionally favourable estimate when LVEF was preserved and a directionally unfavourable one when it was reduced. The crucial calibration is that neither subgroup hazard ratio was individually significant, since both confidence intervals crossed one, and the NYHA interaction was not significant at the conventional threshold. So nothing here is proven; it is a signal worth respecting. The main limitation is the small reduced-ejection-fraction subgroup of only 455 participants, which leaves the harm estimate fragile and wide. Industry sponsorship by the manufacturer and the absence of randomisation by LVEF further temper any causal reading. Can I use this with my patients? Cautiously, yes, in the direction of avoiding once-weekly exenatide as a heart-failure strategy in someone with established reduced ejection fraction or advanced symptoms, where agents with proven benefit are preferable. It does not license prescribing exenatide to protect a preserved-ejection-fraction heart. A dedicated prospective trial stratified by ejection fraction would settle what this analysis can only hint at.
References
Neves JS, Leite AR, Mentz RJ, Holman RR, Zannad F, Butler J, Packer M, Ferreira JP. Cardiovascular outcomes with exenatide in type 2 diabetes according to ejection fraction: the EXSCEL trial. Eur J Heart Fail. 2025;27(3):540-551. doi:10.1002/ejhf.3478
