Summary: In a modelling and mediation analysis of the EXSCEL trial, changes in conventional risk factors explained only a modest share of once-weekly exenatide’s cardiovascular benefits and did not mediate its effect on all-cause mortality, suggesting mechanisms beyond glucose, weight, and blood pressure.
PICO Summary
| Element | Detail |
|---|---|
| Population | Participants with type 2 diabetes in the EXSCEL cardiovascular outcomes trial (post-hoc analysis). |
| Intervention | Once-weekly exenatide, with risk-factor trajectories entered into a validated outcomes model. |
| Comparison | Placebo; simulated versus observed relative risk changes, plus mediation analysis. |
| Outcome | Modelled risk-factor changes explained only modest proportions of observed reductions: MACE 29%, all-cause mortality 15%, CV death 18%, stroke 29%, but more for heart-failure hospitalisation (67%) and MI (200%). Changes in HbA1c, blood pressure, heart rate, LDL, triglycerides, and weight up to 1 year did not mediate the mortality benefit. |
Expert Commentary
This is a thoughtful attempt to answer a question that matters across the whole GLP-1 class: when these drugs reduce cardiovascular events, is it simply through the risk factors we can measure, or through something else? The analysis suggests the latter. Conventional risk-factor changes accounted for only a minority of the mortality and MACE signal and did not mediate the all-cause mortality benefit at all, pointing toward mechanisms such as direct vascular, anti-inflammatory, or myocardial effects that our standard markers do not capture. The heart-failure and myocardial-infarction findings are intriguing but must be read cautiously, an explained proportion of 200% for MI signals model instability rather than a clean inference. As a simulation-and-mediation exercise layered on a single trial whose own MACE result was modest, this is hypothesis-generating about mechanism, not a change to indications. Can I use this with my patients? Indirectly. It reinforces that the cardiovascular value of GLP-1 agonists is not merely a function of how much they lower HbA1c or weight, so I should not withhold or discount them in a patient whose surrogate numbers look only modestly improved. The benefit appears to run deeper than the visible metrics.
References
Coleman RL, Adler AI, Mentz RJ, Fudim M, Sattar N, Holman RR. Impact of changes in conventional risk factors induced by once-weekly GLP-1 receptor agonist exenatide on cardiovascular outcomes: an EXSCEL post hoc analysis. Cardiovasc Diabetol. 2025;24(1):347. doi:10.1186/s12933-025-02866-7