Towards precision secondary prevention after myocardial infarction and stroke with glucagon-like peptide-1 receptor agonists

This editorial refers to ‘GLP-1 receptor agonists for secondary prevention after myocardial infarction and stroke in type 2 diabetes: nationwide real-world evidence’, by P. Sedova et al. https://doi.org/10.1093/eurjpc/zwag002. Patients with Type 2 diabetes (T2D) who survive myocardial infarction (MI) or ischaemic stroke represent one of the highest-risk populations in contemporary cardiovascular practice. Despite significant advances in acute management and secondary prevention, these individuals continue to experience excess rates of recurrent cardiovascular events and premature mortality, reflecting both advanced atherosclerotic disease and the systemic metabolic consequences of diabetes.1 Importantly, such patients have been under-represented in major cardiovascular outcome trials, underscoring the importance of high-quality real-world evidence to inform care in this group. Outcomes following stroke are notably heterogeneous and are influenced by neurological disability, frailty, and competing risks, which collectively challenge long-term preventive strategies.2 Against this backdrop of persistently elevated residual risk, glucagon-like peptide-1 receptor agonists (GLP-1RAs) have emerged over the past decade as therapies with cardiovascular benefits extending beyond glycaemic control. Large cardiovascular outcome trials have consistently demonstrated significant reductions in major adverse cardiovascular events (MACEs) among patients with T2D and established atherosclerotic cardiovascular disease.3,4 These data have informed contemporary clinical guidelines, which now recommend GLP-1RAs for cardiovascular risk reduction in this population, independent of baseline glycaemic status.5 What has remained uncertain, however, is whether the cardiovascular benefits of GLP-1RAs observed in randomized trials translate into routine clinical practice among patients recovering from acute MI or ischaemic stroke. Such patients, particularly those with cerebrovascular disease, have been under-represented in cardiovascular outcome trials, and real-world uptake of GLP-1RAs has remained low. Clarifying both the effectiveness and the implementation of these therapies in post-event populations, therefore, represents a critical evidence gap. The nationwide registry analysis by Sedova et al. provides timely and clinically relevant real-world evidence.6 Using linked Czech national health registries, the investigators examined whether initiation of GLP-1RA therapy after MI or ischaemic stroke in patients with T2D was associated with improved cardiovascular outcomes in routine practice. Among more than 300 000 survivors of MI or ischaemic stroke identified between 2015 and 2024, over 100 000 had confirmed T2D. Despite their very high baseline risk, fewer than 2% of MI survivors and ∼0.5% of stroke survivors initiated GLP-1RA therapy within 12 months of the index event. Although uptake increased modestly over time, it remained notably low, particularly among older adults; women were also consistently less likely to receive GLP-1RA therapy after stroke. Among patients who did initiate treatment, GLP-1RA use was associated with ∼30% lower risk of MACE and 40–45% lower risks of both cardiovascular and all-cause mortality in the MI and stroke cohorts (Figure 1). These associations were directionally consistent across multiple sensitivity analyses designed to address timing-related biases. Collectively, these findings suggest that GLP-1RAs may represent a potent addition to secondary prevention strategies following both MI and ischaemic stroke. Glucagon-like peptide-1 receptor agonist use after myocardial infarction or ischaemic stroke in adults with Type 2 diabetes. GLP-1RA, glucagon-like peptide-1 receptor agonist; MACEs, major adverse cardiovascular events. The magnitude of the observed mortality associations, however, warrants careful interpretation. Hazard ratios approaching 0.55 for cardiovascular death are substantially larger than those reported in most randomized cardiovascular outcome trials of GLP-1RAs3,4 and should not be interpreted as causal. Important determinants of both prognosis and treatment selection, such as frailty and recovery after the index event, are not captured in administrative data and are likely to influence who receives therapy. In addition, GLP-1RA prescribing is shaped by weight-related indications and socioeconomic position (SEP), with higher SEP associated with greater treatment access.7 These unmeasured factors make it unlikely that the observed mortality reduction fully reflects a direct treatment effect and suggest that the benefit is likely overestimated. The authors are to be commended for applying several best-practice approaches in observational effectiveness research, including a new-user design, time-dependent exposure modelling, and multiple sensitivity analyses. These methods meaningfully mitigate immortal time bias and strengthen internal validity, approximating key elements of a target-trial framework.8 Nevertheless, important limitations inherent to registry-based analyses remain. The 12-month treatment initiation window, while clinically pragmatic and reflective of real-world prescribing constraints, is analytically permissive. It allows post-event survival, recovery, healthcare engagement, and clinician assessment of prognosis to influence treatment selection. Moreover, several clinical factors that influence both prescribing decisions and prognosis, including glycaemic control, body mass index, smoking status, frailty, cognitive impairment, and functional recovery, are not captured in administrative registries. Importantly, use of GLP-1RA is socially patterned and has expanded rapidly among individuals with higher SEP, driven in part by their weight-lowering effects. As a result, treated patients are more likely to be fitter, and all of these factors are independently associated with better outcomes. This pattern of selective prescribing is likely to inflate observed benefits through healthy-user effects and to overestimate treatment effects in this study. In contrast, restrictive reimbursement criteria favouring patients with poorer metabolic profiles could bias estimates in the opposite direction. However, given the strong age gradient in prescribing and preferential use among younger, fitter patients, bias towards overestimation is likely to predominate. While consistency across sensitivity analyses strengthens internal coherence, these analyses primarily address timing-related biases and cannot fully resolve confounding by underlying prognosis. The study’s utilization findings are as informative as its outcome associations. The notably low uptake of GLP-1RA therapy after non-fatal MI or ischaemic stroke highlights a substantial gap between clinical trial evidence, guideline recommendations, and routine practice.6 At the same time, this gap warrants careful interpretation. In addition to system-level barriers and therapeutic inertia,9 evolving product safety communications and regulatory reviews, particularly those prompted by post-marketing reports, may contribute to clinician and patient caution when considering GLP-1RA and dual glucose-dependent insulinotropic polypeptide (GIP)/GLP-1RA in real-world settings. Nevertheless, the emergence of safety signals, ongoing pharmacovigilance activities, and practical considerations related to tolerability and treatment burden may still influence prescribing decisions, particularly in vulnerable post-MI or post-stroke populations. Importantly, non-initiation may also be clinically appropriate in selected patients, especially following stroke. Beyond neurological disability, factors such as dysphagia, cognitive impairment, frailty, polypharmacy, and limited life expectancy may reasonably temper enthusiasm for GLP-1RA initiation. Additional concerns related to gastrointestinal intolerance, weight loss, dehydration, and the logistical challenges of injectable therapies further complicate decision-making in vulnerable individuals. Cost and access barriers also play an important role; GLP-1RAs are substantially more expensive than older glucose-lowering agents, and evidence shows that cost, insurance status, and structural inequities meaningfully shape access and utilization.7 The consistently lower uptake observed after stroke compared with MI is, therefore, hypothesis generating and warrants focused evaluation rather than assumption. What should clinicians and health systems take from these findings? This study does not establish causality, nor does it mandate routine initiation of GLP-1RA therapy after MI or stroke. Instead, it provides a strong observational signal, combined with notable underuse, that warrants thoughtful, context-sensitive consideration. For clinicians, these data support consideration of GLP-1RA therapy as part of secondary prevention in selected patients with T2D after MI and in carefully chosen stroke survivors. Decisions should remain individualized, balancing potential cardiovascular benefit against treatment burden and competing risks. For researchers and policymakers, the findings underscore the need for more substantial evidence in post-event populations, whether through pragmatic randomized trials, refined target-trial emulations, or mechanistic studies focused specifically on cerebrovascular outcomes.8,10 The study also highlights the importance of aligning reimbursement policies and care pathways with cardiovascular risk rather than glycaemic thresholds alone, if therapies with established cardiometabolic benefit are to reach patients most likely to benefit. Sedova et al. present one of the most comprehensive real-world evaluations of GLP-1RA therapy among patients with T2D surviving MI or ischaemic stroke. Their findings suggest potentially substantial cardiovascular benefit alongside marked underuse in high-risk populations while also warranting caution in causal interpretation. Rather than resolving the role of GLP-1RAs in secondary prevention, this study sharpens the debate and highlights the need for a more individualized, risk-informed approach. Fariba Ahmadizar (Conceptualization [lead], Writing—original draft [lead]) and Joline W.J. Beulens (Writing—review & editing [supporting])