{
  "abstract": "Introduction Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are increasingly used in cardiometabolic disease for their anti-obesity and anti-diabetic effects, showing great potential as agents for cardiovascular (CV) risk modification. However, their real-world CV adverse event (AE) profile relative to other antidiabetic drugs is not well-characterised, particularly when CV events are examined as clinically meaningful phenotype groups and between sexes. This is clinically relevant given the close association between metabolic disease and CV comorbidity, and recognised underrepresentation of sex-specific CV risk in clinical trials.Methods The Food and Drug Administration Adverse Event Reporting System was analysed to identify single-agent AE reports from 2018-2024 of GLP-1RAs and other antidiabetic drugs. Individual AEs were aggregated into clinically meaningful CV phenotypes: arrhythmic, ischaemic, thromboembolic, heart failure, and autonomic dysfunction. Reporting odds ratios (RORs) were calculated to compare GLP-1RAs as a class against other antidiabetics, as well as within-class comparisons between GLP-1RA agents. All analyses were stratified by sex. Hierarchical clustering and heatmap visualisation of log-transformed RORs were used to assess AE phenotype enrichment between drug classes.Results A total of 4,205 single-drug cardiovascular AE reports were identified across GLP-1RAs, SGLT2 inhibitors, DPP-4 inhibitors, metformin, and insulin. GLP-1RAs demonstrated uniformly lower RORs across all CV phenotypes relative to other antidiabetic classes. Hierarchical clustering showed that GLP-1RAs formed a distinct cluster, separate from higher-risk metabolic cohorts like insulin and SGLT2 inhibitors, while DPP-4 inhibitors clustered with metformin ( figure 1). Heatmapping confirmed strong phenotype-specific enrichment among non-GLP-1 drug classes, in contrast to GLP-1Ras’ relative attenuation across all CV phenotypes (figure 2A). Sex-stratified analyses showed conserved phenotype clustering across antidiabetics, with consistently lower GLP-1RA RORs in males and females (figure 2B).Analyses of specific GLP-1RAs revealed minimal differences between agents overall; however, among females, liraglutide was associated with higher arrhythmic reporting, exenatide with autonomic events, and dulaglutide with lower heart failure reporting. No stable phenotype differences were observed among males.Conclusions In real-world pharmacovigilance data, GLP-1RAs show a sex-stable, distinct CV AE reporting phenotype relative to other antidiabetic drug classes. Within-class analyses show similar reporting patterns between agents, with limited sex-specific deviations. Although these observations may partly reflect underlying patient risk and treatment selection rather than drug effects alone, they support the CV safety of GLP-1RAs as preventive cardiometabolic therapies, while highlighting limited, sex-specific differences between individual agents, that may inform personalised CV risk assessment.Abstract 54 Figure 1(A) Hierarchical clustering of antidiabetic drug classes based on cardiovascular (CV) phenotype reporting odds ratios (RORs) was performed using complete linkage based on Euclidean distance, which demonstrated that GLP-1RAs cluster nearest insulin/SGLT2 inhibitors, but remain distinct from metformin/DPP-4 inhibitors. (B) Heatmap of log-transformed cardiovascular reporting odds ratios (RORs) showing how phenotype distributions drive clustering of antidiabetic drug classes. GLP-1 receptor agonists were uniformly attenuated across all cardiovascular phenotypes, while insulin and SGLT2 inhibitors were enriched for thromboembolic and ischaemic phenotypes respectively. DPP-4 inhibitors and metformin were strongly enriched for heart failure and arrhythmic/hemodynamic domains, respectivelyAbstract 54 Figure 2Heatmap of capped log reporting odds ratios (log ROR), comparing each agent with other GLP-1 receptor agonists only. Cells are masked when based on fewer than five reported cases; hierarchical clustering was performed using unmasked log ROR values. (A) Within-class comparison of GLP-1 receptor agonists showed shared cardiovascular reporting profiles. Liraglutide had higher relative reporting of arrhythmia, exenatide modestly higher reporting of hemodynamic/autonomic events, and dulaglutide lower relative reporting of heart failure. No consistent phenotype differences were observed for semaglutide. (B) Sex-stratified analysis of GLP-1 receptor agonists within this class showed differences among females but not males. In females, there was increased arrhythmic reporting in liraglutide, modestly higher autonomic reporting in exenatide, and lower relative reporting of some phenotypes in dulaglutide. No stable cardiovascular phenotype differences were observed among males",
  "authors": [
    {
      "affiliations": [
        "School of Clinical Medicine, University of Cambridge, Cambridge, United Kingdom"
      ],
      "name": "Justine Tin Nok Chan"
    }
  ],
  "title": "54 Cardiovascular safety of glucagon-like peptide-1 receptor agonists in real-world practice: a comparative phenotype-based analysis",
  "uid": "641de05f-9fbd-5a58-9ee0-896faa834dfc"
}